1 acpi= [HW,ACPI,X86,ARM64]
2 Advanced Configuration and Power Interface
3 Format: { force | on | off | strict | noirq | rsdt |
5 force -- enable ACPI if default was off
6 on -- enable ACPI but allow fallback to DT [arm64]
7 off -- disable ACPI if default was on
8 noirq -- do not use ACPI for IRQ routing
9 strict -- Be less tolerant of platforms that are not
10 strictly ACPI specification compliant.
11 rsdt -- prefer RSDT over (default) XSDT
12 copy_dsdt -- copy DSDT to memory
13 For ARM64, ONLY "acpi=off", "acpi=on" or "acpi=force"
16 See also Documentation/power/runtime_pm.rst, pci=noacpi
18 acpi_apic_instance= [ACPI, IOAPIC]
20 2: use 2nd APIC table, if available
21 1,0: use 1st APIC table
24 acpi_backlight= [HW,ACPI]
25 { vendor | video | native | none }
26 If set to vendor, prefer vendor-specific driver
27 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
28 of the ACPI video.ko driver.
29 If set to video, use the ACPI video.ko driver.
30 If set to native, use the device's native backlight mode.
31 If set to none, disable the ACPI backlight interface.
33 acpi_force_32bit_fadt_addr
34 force FADT to use 32 bit addresses rather than the
35 64 bit X_* addresses. Some firmware have broken 64
36 bit addresses for force ACPI ignore these and use
37 the older legacy 32 bit addresses.
39 acpica_no_return_repair [HW, ACPI]
40 Disable AML predefined validation mechanism
41 This mechanism can repair the evaluation result to make
42 the return objects more ACPI specification compliant.
43 This option is useful for developers to identify the
44 root cause of an AML interpreter issue when the issue
45 has something to do with the repair mechanism.
47 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
48 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
50 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
51 debug output. Bits in debug_layer correspond to a
52 _COMPONENT in an ACPI source file, e.g.,
53 #define _COMPONENT ACPI_EVENTS
54 Bits in debug_level correspond to a level in
55 ACPI_DEBUG_PRINT statements, e.g.,
56 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
57 The debug_level mask defaults to "info". See
58 Documentation/firmware-guide/acpi/debug.rst for more information about
59 debug layers and levels.
61 Enable processor driver info messages:
62 acpi.debug_layer=0x20000000
63 Enable AML "Debug" output, i.e., stores to the Debug
64 object while interpreting AML:
65 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
66 Enable all messages related to ACPI hardware:
67 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
69 Some values produce so much output that the system is
70 unusable. The "log_buf_len" parameter may be useful
71 if you need to capture more output.
73 acpi_enforce_resources= [ACPI]
75 Check for resource conflicts between native drivers
76 and ACPI OperationRegions (SystemIO and SystemMemory
77 only). IO ports and memory declared in ACPI might be
78 used by the ACPI subsystem in arbitrary AML code and
79 can interfere with legacy drivers.
80 strict (default): access to resources claimed by ACPI
81 is denied; legacy drivers trying to access reserved
82 resources will fail to bind to device using them.
83 lax: access to resources claimed by ACPI is allowed;
84 legacy drivers trying to access reserved resources
85 will bind successfully but a warning message is logged.
86 no: ACPI OperationRegions are not marked as reserved,
87 no further checks are performed.
89 acpi_force_table_verification [HW,ACPI]
90 Enable table checksum verification during early stage.
91 By default, this is disabled due to x86 early mapping
94 acpi_irq_balance [HW,ACPI]
95 ACPI will balance active IRQs
98 acpi_irq_nobalance [HW,ACPI]
99 ACPI will not move active IRQs (default)
102 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
103 Format: <irq>,<irq>...
105 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
107 Format: <irq>,<irq>...
109 acpi_mask_gpe= [HW,ACPI]
110 Due to the existence of _Lxx/_Exx, some GPEs triggered
111 by unsupported hardware/firmware features can result in
112 GPE floodings that cannot be automatically disabled by
114 This facility can be used to prevent such uncontrolled
116 Format: <byte> or <bitmap-list>
118 acpi_no_auto_serialize [HW,ACPI]
119 Disable auto-serialization of AML methods
120 AML control methods that contain the opcodes to create
121 named objects will be marked as "Serialized" by the
122 auto-serialization feature.
123 This feature is enabled by default.
124 This option allows to turn off the feature.
126 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
129 acpi_no_static_ssdt [HW,ACPI]
130 Disable installation of static SSDTs at early boot time
131 By default, SSDTs contained in the RSDT/XSDT will be
132 installed automatically and they will appear under
133 /sys/firmware/acpi/tables.
134 This option turns off this feature.
135 Note that specifying this option does not affect
136 dynamic table installation which will install SSDT
137 tables to /sys/firmware/acpi/tables/dynamic.
139 acpi_no_watchdog [HW,ACPI,WDT]
140 Ignore the ACPI-based watchdog interface (WDAT) and let
141 a native driver control the watchdog device instead.
143 acpi_rsdp= [ACPI,EFI,KEXEC]
144 Pass the RSDP address to the kernel, mostly used
145 on machines running EFI runtime service to boot the
146 second kernel for kdump.
148 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
149 Format: To spoof as Windows 98: ="Microsoft Windows"
151 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
152 of 2 which is mandated by ACPI 6) as the supported ACPI
153 specification revision (when using this switch, it may
154 be necessary to carry out a cold reboot _twice_ in a
155 row to make it take effect on the platform firmware).
157 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
158 acpi_osi="string1" # add string1
159 acpi_osi="!string2" # remove string2
160 acpi_osi=!* # remove all strings
161 acpi_osi=! # disable all built-in OS vendor
163 acpi_osi=!! # enable all built-in OS vendor
165 acpi_osi= # disable all strings
167 'acpi_osi=!' can be used in combination with single or
168 multiple 'acpi_osi="string1"' to support specific OS
169 vendor string(s). Note that such command can only
170 affect the default state of the OS vendor strings, thus
171 it cannot affect the default state of the feature group
172 strings and the current state of the OS vendor strings,
173 specifying it multiple times through kernel command line
174 is meaningless. This command is useful when one do not
175 care about the state of the feature group strings which
176 should be controlled by the OSPM.
178 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
179 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
180 can make '_OSI("Windows 2000")' TRUE.
182 'acpi_osi=' cannot be used in combination with other
183 'acpi_osi=' command lines, the _OSI method will not
184 exist in the ACPI namespace. NOTE that such command can
185 only affect the _OSI support state, thus specifying it
186 multiple times through kernel command line is also
189 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
192 'acpi_osi=!*' can be used in combination with single or
193 multiple 'acpi_osi="string1"' to support specific
194 string(s). Note that such command can affect the
195 current state of both the OS vendor strings and the
196 feature group strings, thus specifying it multiple times
197 through kernel command line is meaningful. But it may
198 still not able to affect the final state of a string if
199 there are quirks related to this string. This command
200 is useful when one want to control the state of the
201 feature group strings to debug BIOS issues related to
204 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
205 '_OSI("Module Device")' FALSE.
206 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
207 '_OSI("Module Device")' TRUE.
208 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
210 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
212 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
213 they all will make '_OSI("Windows 2000")' TRUE.
216 Override the pmtimer bug detection: force the kernel
217 to assume that this machine's pmtimer latches its value
218 and always returns good values.
220 acpi_sci= [HW,ACPI] ACPI System Control Interrupt trigger mode
221 Format: { level | edge | high | low }
223 acpi_skip_timer_override [HW,ACPI]
224 Recognize and ignore IRQ0/pin2 Interrupt Override.
225 For broken nForce2 BIOS resulting in XT-PIC timer.
227 acpi_sleep= [HW,ACPI] Sleep options
228 Format: { s3_bios, s3_mode, s3_beep, s4_nohwsig,
229 old_ordering, nonvs, sci_force_enable, nobl }
230 See Documentation/power/video.rst for information on
232 s3_beep is for debugging; it makes the PC's speaker beep
233 as soon as the kernel's real-mode entry point is called.
234 s4_nohwsig prevents ACPI hardware signature from being
235 used during resume from hibernation.
236 old_ordering causes the ACPI 1.0 ordering of the _PTS
237 control method, with respect to putting devices into
238 low power states, to be enforced (the ACPI 2.0 ordering
239 of _PTS is used by default).
240 nonvs prevents the kernel from saving/restoring the
241 ACPI NVS memory during suspend/hibernation and resume.
242 sci_force_enable causes the kernel to set SCI_EN directly
243 on resume from S1/S3 (which is against the ACPI spec,
244 but some broken systems don't work without it).
245 nobl causes the internal blacklist of systems known to
246 behave incorrectly in some ways with respect to system
247 suspend and resume to be ignored (use wisely).
249 acpi_use_timer_override [HW,ACPI]
250 Use timer override. For some broken Nvidia NF5 boards
251 that require a timer override, but don't have HPET
253 add_efi_memmap [EFI; X86] Include EFI memory map in
254 kernel's map of available physical RAM.
257 { off | try_unsupported }
258 off: disable AGP support
259 try_unsupported: try to drive unsupported chipsets
260 (may crash computer or cause data corruption)
263 See Documentation/sound/alsa-configuration.rst
266 Allow the default userspace alignment fault handler
267 behaviour to be specified. Bit 0 enables warnings,
268 bit 1 enables fixups, and bit 2 sends a segfault.
270 align_va_addr= [X86-64]
271 Align virtual addresses by clearing slice [14:12] when
272 allocating a VMA at process creation time. This option
273 gives you up to 3% performance improvement on AMD F15h
274 machines (where it is enabled by default) for a
275 CPU-intensive style benchmark, and it can vary highly in
276 a microbenchmark depending on workload and compiler.
278 32: only for 32-bit processes
279 64: only for 64-bit processes
280 on: enable for both 32- and 64-bit processes
281 off: disable for both 32- and 64-bit processes
283 alloc_snapshot [FTRACE]
284 Allocate the ftrace snapshot buffer on boot up when the
285 main buffer is allocated. This is handy if debugging
286 and you need to use tracing_snapshot() on boot up, and
287 do not want to use tracing_snapshot_alloc() as it needs
288 to be done where GFP_KERNEL allocations are allowed.
290 allow_mismatched_32bit_el0 [ARM64]
291 Allow execve() of 32-bit applications and setting of the
292 PER_LINUX32 personality on systems where only a strict
293 subset of the CPUs support 32-bit EL0. When this
294 parameter is present, the set of CPUs supporting 32-bit
295 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
296 and hot-unplug operations may be restricted.
298 See Documentation/arm64/asymmetric-32bit.rst for more
301 amd_iommu= [HW,X86-64]
302 Pass parameters to the AMD IOMMU driver in the system.
304 fullflush - Deprecated, equivalent to iommu.strict=1
305 off - do not initialize any AMD IOMMU found in
307 force_isolation - Force device isolation for all
308 devices. The IOMMU driver is not
309 allowed anymore to lift isolation
310 requirements as needed. This option
311 does not override iommu=pt
312 force_enable - Force enable the IOMMU on platforms known
313 to be buggy with IOMMU enabled. Use this
316 amd_iommu_dump= [HW,X86-64]
317 Enable AMD IOMMU driver option to dump the ACPI table
318 for AMD IOMMU. With this option enabled, AMD IOMMU
319 driver will print ACPI tables for AMD IOMMU during
320 IOMMU initialization.
322 amd_iommu_intr= [HW,X86-64]
323 Specifies one of the following AMD IOMMU interrupt
325 legacy - Use legacy interrupt remapping mode.
326 vapic - Use virtual APIC mode, which allows IOMMU
327 to inject interrupts directly into guest.
328 This mode requires kvm-amd.avic=1.
329 (Default when IOMMU HW support is present.)
331 amijoy.map= [HW,JOY] Amiga joystick support
332 Map of devices attached to JOY0DAT and JOY1DAT
334 See also Documentation/input/joydev/joystick.rst
336 analog.map= [HW,JOY] Analog joystick and gamepad support
337 Specifies type or capabilities of an analog joystick
338 connected to one of 16 gameports
339 Format: <type1>,<type2>,..<type16>
342 Power management functions (SPARCstation-4/5 + deriv.)
344 Disable APC CPU standby support. SPARCstation-Fox does
345 not play well with APC CPU idle - disable it if you have
346 APC and your system crashes randomly.
348 apic= [APIC,X86] Advanced Programmable Interrupt Controller
349 Change the output verbosity while booting
350 Format: { quiet (default) | verbose | debug }
351 Change the amount of debugging information output
352 when initialising the APIC and IO-APIC components.
353 For X86-32, this can also be used to specify an APIC
355 Format: apic=driver_name
356 Examples: apic=bigsmp
358 apic_extnmi= [APIC,X86] External NMI delivery setting
359 Format: { bsp (default) | all | none }
360 bsp: External NMI is delivered only to CPU 0
361 all: External NMIs are broadcast to all CPUs as a
363 none: External NMI is masked for all CPUs. This is
364 useful so that a dump capture kernel won't be
368 See Documentation/networking/ipv6.rst.
370 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
371 Limit apic dumping. The parameter defines the maximal
372 number of local apics being dumped. Also it is possible
373 to set it to "all" by meaning -- no limit here.
374 Format: { 1 (default) | 2 | ... | all }.
375 The parameter valid if only apic=debug or
376 apic=verbose is specified.
377 Example: apic=debug show_lapic=all
379 apm= [APM] Advanced Power Management
380 See header of arch/x86/kernel/apm_32.c.
382 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
383 Format: <io>,<irq>,<nodeID>
385 arm64.nobti [ARM64] Unconditionally disable Branch Target
386 Identification support
388 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
391 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
396 atarimouse= [HW,MOUSE] Atari Mouse
398 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
399 EzKey and similar keyboards
401 atkbd.reset= [HW] Reset keyboard during initialization
403 atkbd.set= [HW] Select keyboard code set
404 Format: <int> (2 = AT (default), 3 = PS/2)
406 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
409 atkbd.softraw= [HW] Choose between synthetic and real raw mode
410 Format: <bool> (0 = real, 1 = synthetic (default))
412 atkbd.softrepeat= [HW]
413 Use software keyboard repeat
415 audit= [KNL] Enable the audit sub-system
416 Format: { "0" | "1" | "off" | "on" }
417 0 | off - kernel audit is disabled and can not be
418 enabled until the next reboot
419 unset - kernel audit is initialized but disabled and
420 will be fully enabled by the userspace auditd.
421 1 | on - kernel audit is initialized and partially
422 enabled, storing at most audit_backlog_limit
423 messages in RAM until it is fully enabled by the
427 audit_backlog_limit= [KNL] Set the audit queue size limit.
428 Format: <int> (must be >=0)
431 bau= [X86_UV] Enable the BAU on SGI UV. The default
432 behavior is to disable the BAU (i.e. bau=0).
433 Format: { "0" | "1" }
436 unset - Disable the BAU.
438 baycom_epp= [HW,AX25]
441 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
443 See header of drivers/net/hamradio/baycom_par.c.
445 baycom_ser_fdx= [HW,AX25]
446 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
447 Format: <io>,<irq>,<mode>[,<baud>]
448 See header of drivers/net/hamradio/baycom_ser_fdx.c.
450 baycom_ser_hdx= [HW,AX25]
451 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
452 Format: <io>,<irq>,<mode>
453 See header of drivers/net/hamradio/baycom_ser_hdx.c.
455 blkdevparts= Manual partition parsing of block device(s) for
456 embedded devices based on command line input.
457 See Documentation/block/cmdline-partition.rst
459 boot_delay= Milliseconds to delay each printk during boot.
460 Values larger than 10 seconds (10000) are changed to
465 Extended command line options can be added to an initrd
466 and this will cause the kernel to look for it.
468 See Documentation/admin-guide/bootconfig.rst
471 Disable BERT OS support on buggy BIOSes.
473 bgrt_disable [ACPI][X86]
474 Disable BGRT to avoid flickering OEM logo.
476 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
477 bttv.radio= Most important insmod options are available as
479 bttv.pll= See Documentation/admin-guide/media/bttv.rst
482 bulk_remove=off [PPC] This parameter disables the use of the pSeries
483 firmware feature for flushing multiple hpte entries
486 c101= [NET] Moxa C101 synchronous serial card
488 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
489 Sometimes CPU hardware bugs make them report the cache
490 size incorrectly. The kernel will attempt work arounds
491 to fix known problems, but for some CPUs it is not
492 possible to determine what the correct size should be.
493 This option provides an override for these situations.
496 [NET] Specifies amount of time (in seconds) that
497 the kernel should wait for a network carrier. By default
498 it waits 120 seconds.
500 ca_keys= [KEYS] This parameter identifies a specific key(s) on
501 the system trusted keyring to be used for certificate
503 format: { id:<keyid> | builtin }
505 cca= [MIPS] Override the kernel pages' cache coherency
506 algorithm. Accepted values range from 0 to 7
507 inclusive. See arch/mips/include/asm/pgtable-bits.h
508 for platform specific values (SB1, Loongson3 and
511 ccw_timeout_log [S390]
512 See Documentation/s390/common_io.rst for details.
514 cgroup_disable= [KNL] Disable a particular controller or optional feature
515 Format: {name of the controller(s) or feature(s) to disable}
516 The effects of cgroup_disable=foo are:
517 - foo isn't auto-mounted if you mount all cgroups in
519 - foo isn't visible as an individually mountable
521 - if foo is an optional feature then the feature is
522 disabled and corresponding cgroup files are not
524 {Currently only "memory" controller deal with this and
525 cut the overhead, others just disable the usage. So
526 only cgroup_disable=memory is actually worthy}
527 Specifying "pressure" disables per-cgroup pressure
528 stall information accounting feature
530 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
531 Format: { { controller | "all" | "named" }
532 [,{ controller | "all" | "named" }...] }
533 Like cgroup_disable, but only applies to cgroup v1;
534 the blacklisted controllers remain available in cgroup2.
535 "all" blacklists all controllers and "named" disables
536 named mounts. Specifying both "all" and "named" disables
539 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
541 nosocket -- Disable socket memory accounting.
542 nokmem -- Disable kernel memory accounting.
544 checkreqprot [SELINUX] Set initial checkreqprot flag value.
545 Format: { "0" | "1" }
546 See security/selinux/Kconfig help text.
547 0 -- check protection applied by kernel (includes
548 any implied execute protection).
549 1 -- check protection requested by application.
550 Default value is set via a kernel config option.
551 Value can be changed at runtime via
552 /sys/fs/selinux/checkreqprot.
553 Setting checkreqprot to 1 is deprecated.
556 See Documentation/s390/common_io.rst for details.
559 Prevents the clock framework from automatically gating
560 clocks that have not been explicitly enabled by a Linux
561 device driver but are enabled in hardware at reset or
562 by the bootloader/firmware. Note that this does not
563 force such clocks to be always-on nor does it reserve
564 those clocks in any way. This parameter is useful for
565 debug and development, but should not be needed on a
566 platform with proper driver support. For more
567 information, see Documentation/driver-api/clk.rst.
569 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
571 Forces specified clocksource (if available) to be used
572 when calculating gettimeofday(). If specified
573 clocksource is not available, it defaults to PIT.
574 Format: { pit | tsc | cyclone | pmtmr }
576 clocksource= Override the default clocksource
578 Override the default clocksource and use the clocksource
579 with the name specified.
580 Some clocksource names to choose from, depending on
582 [all] jiffies (this is the base, fallback clocksource)
584 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
585 pxa_timer,timer3,32k_counter,timer0_1
586 [X86-32] pit,hpet,tsc;
587 scx200_hrt on Geode; cyclone on IBM x440
595 clocksource.arm_arch_timer.evtstrm=
598 Enable/disable the eventstream feature of the ARM
599 architected timer so that code using WFE-based polling
600 loops can be debugged more effectively on production
603 clocksource.max_cswd_read_retries= [KNL]
604 Number of clocksource_watchdog() retries due to
605 external delays before the clock will be marked
606 unstable. Defaults to three retries, that is,
607 four attempts to read the clock under test.
609 clocksource.verify_n_cpus= [KNL]
610 Limit the number of CPUs checked for clocksources
611 marked with CLOCK_SOURCE_VERIFY_PERCPU that
612 are marked unstable due to excessive skew.
613 A negative value says to check all CPUs, while
614 zero says not to check any. Values larger than
615 nr_cpu_ids are silently truncated to nr_cpu_ids.
616 The actual CPUs are chosen randomly, with
617 no replacement if the same CPU is chosen twice.
619 clocksource-wdtest.holdoff= [KNL]
620 Set the time in seconds that the clocksource
621 watchdog test waits before commencing its tests.
622 Defaults to zero when built as a module and to
623 10 seconds when built into the kernel.
625 clearcpuid=BITNUM[,BITNUM...] [X86]
626 Disable CPUID feature X for the kernel. See
627 arch/x86/include/asm/cpufeatures.h for the valid bit
628 numbers. Note the Linux specific bits are not necessarily
629 stable over kernel options, but the vendor specific
631 Also note that user programs calling CPUID directly
632 or using the feature without checking anything
633 will still see it. This just prevents it from
634 being used by the kernel or shown in /proc/cpuinfo.
635 Also note the kernel might malfunction if you disable
638 cma=nn[MG]@[start[MG][-end[MG]]]
640 Sets the size of kernel global memory area for
641 contiguous memory allocations and optionally the
642 placement constraint by the physical address range of
643 memory allocations. A value of 0 disables CMA
644 altogether. For more information, see
645 kernel/dma/contiguous.c
649 Sets the size of kernel per-numa memory area for
650 contiguous memory allocations. A value of 0 disables
651 per-numa CMA altogether. And If this option is not
652 specificed, the default value is 0.
653 With per-numa CMA enabled, DMA users on node nid will
654 first try to allocate buffer from the pernuma area
655 which is located in node nid, if the allocation fails,
656 they will fallback to the global default memory area.
658 cmo_free_hint= [PPC] Format: { yes | no }
659 Specify whether pages are marked as being inactive
660 when they are freed. This is used in CMO environments
661 to determine OS memory pressure for page stealing by
665 coherent_pool=nn[KMG] [ARM,KNL]
666 Sets the size of memory pool for coherent, atomic dma
667 allocations, by default set to 256K.
669 com20020= [HW,NET] ARCnet - COM20020 chipset
671 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
673 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
677 ARCnet - COM90xx chipset (memory-mapped buffers)
678 Format: <io>[,<irq>[,<memstart>]]
680 condev= [HW,S390] console device
683 console= [KNL] Output console device and options.
685 tty<n> Use the virtual console device <n>.
689 Use the specified serial port. The options are of
690 the form "bbbbpnf", where "bbbb" is the baud rate,
691 "p" is parity ("n", "o", or "e"), "n" is number of
692 bits, and "f" is flow control ("r" for RTS or
693 omit it). Default is "9600n8".
695 See Documentation/admin-guide/serial-console.rst for more
697 Documentation/networking/netconsole.rst for an
700 uart[8250],io,<addr>[,options]
701 uart[8250],mmio,<addr>[,options]
702 uart[8250],mmio16,<addr>[,options]
703 uart[8250],mmio32,<addr>[,options]
704 uart[8250],0x<addr>[,options]
705 Start an early, polled-mode console on the 8250/16550
706 UART at the specified I/O port or MMIO address,
707 switching to the matching ttyS device later.
708 MMIO inter-register address stride is either 8-bit
709 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
710 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
711 to be equivalent to 'mmio'. 'options' are specified in
712 the same format described for ttyS above; if unspecified,
713 the h/w is not re-initialized.
715 hvc<n> Use the hypervisor console device <n>. This is for
716 both Xen and PowerPC hypervisors.
718 If the device connected to the port is not a TTY but a braille
719 device, prepend "brl," before the device type, for instance
721 For now, only VisioBraille is supported.
724 [KNL] Change console messages format
726 By default we print messages on consoles in
727 "[time stamp] text\n" format (time stamp may not be
728 printed, depending on CONFIG_PRINTK_TIME or
729 `printk_time' param).
731 Switch to syslog format: "<%u>[time stamp] text\n"
732 IOW, each message will have a facility and loglevel
733 prefix. The format is similar to one used by syslog()
734 syscall, or to executing "dmesg -S --raw" or to reading
737 consoleblank= [KNL] The console blank (screen saver) timeout in
738 seconds. A value of 0 disables the blank timer.
742 [KNL] Change the default value for
743 /proc/<pid>/coredump_filter.
744 See also Documentation/filesystems/proc.rst.
746 coresight_cpu_debug.enable
749 Enable/disable the CPU sampling based debugging.
750 0: default value, disable debugging
751 1: enable debugging at boot time
753 cpuidle.off=1 [CPU_IDLE]
754 disable the cpuidle sub-system
757 [CPU_IDLE] Name of the cpuidle governor to use.
759 cpufreq.off=1 [CPU_FREQ]
760 disable the cpufreq sub-system
762 cpufreq.default_governor=
763 [CPU_FREQ] Name of the default cpufreq governor or
764 policy to use. This governor must be registered in the
765 kernel before the cpufreq driver probes.
768 [X86] Delay for N microsec between assert and de-assert
769 of APIC INIT to start processors. This delay occurs
770 on every CPU online, such as boot, and resume from suspend.
773 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
775 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
777 crashkernel=size[KMG][@offset[KMG]]
778 [KNL] Using kexec, Linux can switch to a 'crash kernel'
779 upon panic. This parameter reserves the physical
780 memory region [offset, offset + size] for that kernel
781 image. If '@offset' is omitted, then a suitable offset
782 is selected automatically.
783 [KNL, X86-64] Select a region under 4G first, and
784 fall back to reserve region above 4G when '@offset'
785 hasn't been specified.
786 See Documentation/admin-guide/kdump/kdump.rst for further details.
788 crashkernel=range1:size1[,range2:size2,...][@offset]
789 [KNL] Same as above, but depends on the memory
790 in the running system. The syntax of range is
791 start-[end] where start and end are both
792 a memory unit (amount[KMG]). See also
793 Documentation/admin-guide/kdump/kdump.rst for an example.
795 crashkernel=size[KMG],high
796 [KNL, X86-64] range could be above 4G. Allow kernel
797 to allocate physical memory region from top, so could
798 be above 4G if system have more than 4G ram installed.
799 Otherwise memory region will be allocated below 4G, if
801 It will be ignored if crashkernel=X is specified.
802 crashkernel=size[KMG],low
803 [KNL, X86-64] range under 4G. When crashkernel=X,high
804 is passed, kernel could allocate physical memory region
805 above 4G, that cause second kernel crash on system
806 that require some amount of low memory, e.g. swiotlb
807 requires at least 64M+32K low memory, also enough extra
808 low memory is needed to make sure DMA buffers for 32-bit
809 devices won't run out. Kernel would try to allocate at
810 at least 256M below 4G automatically.
811 This one let user to specify own low range under 4G
812 for second kernel instead.
813 0: to disable low allocation.
814 It will be ignored when crashkernel=X,high is not used
815 or memory reserved is below 4G.
818 [KNL] Disable crypto self-tests
823 cs89x0_media= [HW,NET]
824 Format: { rj45 | aui | bnc }
826 csdlock_debug= [KNL] Enable debug add-ons of cross-CPU function call
827 handling. When switched on, additional debug data is
828 printed to the console in case a hanging CPU is
829 detected, and that CPU is pinged again in order to try
830 to resolve the hang situation.
831 0: disable csdlock debugging (default)
832 1: enable basic csdlock debugging (minor impact)
833 ext: enable extended csdlock debugging (more impact,
837 See header of drivers/s390/block/dasd_devmap.c.
839 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
840 (one device per port)
841 Format: <port#>,<type>
842 See also Documentation/input/devices/joystick-parport.rst
844 ddebug_query= [KNL,DYNAMIC_DEBUG] Enable debug messages at early boot
846 Documentation/admin-guide/dynamic-debug-howto.rst for
847 details. Deprecated, see dyndbg.
849 debug [KNL] Enable kernel debugging (events log level).
852 [KNL] Enable printing [hashed] pointers early in the
853 boot sequence. If enabled, we use a weak hash instead
854 of siphash to hash pointers. Use this option if you are
855 seeing instances of '(___ptrval___)') and need to see a
856 value (hashed pointer) instead. Cryptographically
857 insecure, please do not use on production kernels.
860 [KNL] verbose locking self-tests
862 Print debugging info while doing the locking API
864 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
865 (no extra messages), setting it to -1 (all bits set)
866 will print _a_lot_ more information - normally only
867 useful to lockdep developers.
869 debug_objects [KNL] Enable object debugging
872 [KNL] Disable object debugging
874 debug_guardpage_minorder=
875 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this
876 parameter allows control of the order of pages that will
877 be intentionally kept free (and hence protected) by the
878 buddy allocator. Bigger value increase the probability
879 of catching random memory corruption, but reduce the
880 amount of memory for normal system use. The maximum
881 possible value is MAX_ORDER/2. Setting this parameter
882 to 1 or 2 should be enough to identify most random
883 memory corruption problems caused by bugs in kernel or
884 driver code when a CPU writes to (or reads from) a
885 random memory location. Note that there exists a class
886 of memory corruptions problems caused by buggy H/W or
887 F/W or by drivers badly programing DMA (basically when
888 memory is written at bus level and the CPU MMU is
889 bypassed) which are not detectable by
890 CONFIG_DEBUG_PAGEALLOC, hence this option will not help
891 tracking down these problems.
894 [KNL] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
895 enables the feature at boot time. By default, it is
896 disabled and the system will work mostly the same as a
897 kernel built without CONFIG_DEBUG_PAGEALLOC.
898 Note: to get most of debug_pagealloc error reports, it's
899 useful to also enable the page_owner functionality.
900 on: enable the feature
902 debugfs= [KNL] This parameter enables what is exposed to userspace
903 and debugfs internal clients.
904 Format: { on, no-mount, off }
905 on: All functions are enabled.
907 Filesystem is not registered but kernel clients can
908 access APIs and a crashkernel can be used to read
909 its content. There is nothing to mount.
910 off: Filesystem is not registered and clients
911 get a -EPERM as result when trying to register files
912 or directories within debugfs.
913 This is equivalent of the runtime functionality if
914 debugfs was not enabled in the kernel at all.
915 Default value is set in build-time with a kernel configuration.
917 debugpat [X86] Enable PAT debugging
920 [HW] The size of the default HugeTLB page. This is
921 the size represented by the legacy /proc/ hugepages
922 APIs. In addition, this is the default hugetlb size
923 used for shmget(), mmap() and mounting hugetlbfs
924 filesystems. If not specified, defaults to the
925 architecture's default huge page size. Huge page
926 sizes are architecture dependent. See also
927 Documentation/admin-guide/mm/hugetlbpage.rst.
930 deferred_probe_timeout=
931 [KNL] Debugging option to set a timeout in seconds for
932 deferred probe to give up waiting on dependencies to
933 probe. Only specific dependencies (subsystems or
934 drivers) that have opted in will be ignored. A timeout of 0
935 will timeout at the end of initcalls. This option will also
936 dump out devices still on the deferred probe list after
940 Format: { on | off | def_only | inf_only | always }
941 on: s390 zlib hardware support for compression on
942 level 1 and decompression (default)
943 off: No s390 zlib hardware support
944 def_only: s390 zlib hardware support for deflate
945 only (compression on level 1)
946 inf_only: s390 zlib hardware support for inflate
948 always: Same as 'on' but ignores the selected compression
949 level always using hardware support (used for debugging)
952 Set number of hash buckets for dentry cache.
954 disable_1tb_segments [PPC]
955 Disables the use of 1TB hash page table segments. This
956 causes the kernel to fall back to 256MB segments which
957 can be useful when debugging issues that require an SLB
961 Limits the number of kernel SLB entries, and flushes
962 them frequently to increase the rate of SLB faults
966 See Documentation/networking/ipv6.rst.
969 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
970 hardening is enabled for this boot. Hardened
971 usercopy checking is used to protect the kernel
972 from reading or writing beyond known memory
973 allocation boundaries as a proactive defense
974 against bounds-checking flaws in the kernel's
975 copy_to_user()/copy_from_user() interface.
976 on Perform hardened usercopy checks (default).
977 off Disable hardened usercopy checks.
980 Disable RADIX MMU mode on POWER9
982 radix_hcall_invalidate=on [PPC/PSERIES]
983 Disable RADIX GTSE feature and use hcall for TLB
987 Disable TLBIE instruction. Currently does not work
988 with KVM, with HASH MMU, or with coherent accelerators.
990 disable_cpu_apicid= [X86,APIC,SMP]
992 The number of initial APIC ID for the
993 corresponding CPU to be disabled at boot,
994 mostly used for the kdump 2nd kernel to
995 disable BSP to wake up multiple CPUs without
996 causing system reset or hang due to sending
999 disable_ddw [PPC/PSERIES]
1000 Disable Dynamic DMA Window support. Use this
1001 to workaround buggy firmware.
1003 disable_ipv6= [IPV6]
1004 See Documentation/networking/ipv6.rst.
1006 disable_mtrr_cleanup [X86]
1007 The kernel tries to adjust MTRR layout from continuous
1008 to discrete, to make X server driver able to add WB
1009 entry later. This parameter disables that.
1011 disable_mtrr_trim [X86, Intel and AMD only]
1012 By default the kernel will trim any uncacheable
1013 memory out of your available memory pool based on
1014 MTRR settings. This parameter disables that behavior,
1015 possibly causing your machine to run very slowly.
1017 disable_timer_pin_1 [X86]
1018 Disable PIN 1 of APIC timer
1019 Can be useful to work around chipset bugs.
1021 dis_ucode_ldr [X86] Disable the microcode loader.
1023 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1024 this option disables the debugging code at boot.
1026 dma_debug_entries=<number>
1027 This option allows to tune the number of preallocated
1028 entries for DMA-API debugging code. One entry is
1029 required per DMA-API allocation. Use this if the
1030 DMA-API debugging code disables itself because the
1031 architectural default is too low.
1033 dma_debug_driver=<driver_name>
1034 With this option the DMA-API debugging driver
1035 filter feature can be enabled at boot time. Just
1036 pass the driver to filter for as the parameter.
1037 The filter can be disabled or changed to another
1038 driver later using sysfs.
1040 driver_async_probe= [KNL]
1041 List of driver names to be probed asynchronously.
1042 Format: <driver_name1>,<driver_name2>...
1044 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1045 Broken monitors, graphic adapters, KVMs and EDIDless
1046 panels may send no or incorrect EDID data sets.
1047 This parameter allows to specify an EDID data sets
1048 in the /lib/firmware directory that are used instead.
1049 Generic built-in EDID data sets are used, if one of
1050 edid/1024x768.bin, edid/1280x1024.bin,
1051 edid/1680x1050.bin, or edid/1920x1080.bin is given
1052 and no file with the same name exists. Details and
1053 instructions how to build your own EDID data are
1054 available in Documentation/admin-guide/edid.rst. An EDID
1055 data set will only be used for a particular connector,
1056 if its name and a colon are prepended to the EDID
1057 name. Each connector may use a unique EDID data
1058 set by separating the files with a comma. An EDID
1059 data set with no connector name will be used for
1060 any connectors not explicitly specified.
1065 Format: {"off" | "known"}
1066 Control how the dt_cpu_ftrs device-tree binding is
1067 used for CPU feature discovery and setup (if it
1069 off: Do not use it, fall back to legacy cpu table.
1070 known: Do not pass through unknown features to guests
1071 or userspace, only those that the kernel is aware of.
1073 dump_apple_properties [X86]
1074 Dump name and content of EFI device properties on
1075 x86 Macs. Useful for driver authors to determine
1076 what data is available or for reverse-engineering.
1078 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1079 <module>.dyndbg[="val"]
1080 Enable debug messages at boot time. See
1081 Documentation/admin-guide/dynamic-debug-howto.rst
1084 nopku [X86] Disable Memory Protection Keys CPU feature found
1087 <module>.async_probe [KNL]
1088 Enable asynchronous probe on this module.
1090 early_ioremap_debug [KNL]
1091 Enable debug messages in early_ioremap support. This
1092 is useful for tracking down temporary early mappings
1093 which are not unmapped.
1095 earlycon= [KNL] Output early console device and options.
1097 When used with no options, the early console is
1098 determined by stdout-path property in device tree's
1099 chosen node or the ACPI SPCR table if supported by
1102 cdns,<addr>[,options]
1103 Start an early, polled-mode console on a Cadence
1104 (xuartps) serial port at the specified address. Only
1105 supported option is baud rate. If baud rate is not
1106 specified, the serial port must already be setup and
1109 uart[8250],io,<addr>[,options]
1110 uart[8250],mmio,<addr>[,options]
1111 uart[8250],mmio32,<addr>[,options]
1112 uart[8250],mmio32be,<addr>[,options]
1113 uart[8250],0x<addr>[,options]
1114 Start an early, polled-mode console on the 8250/16550
1115 UART at the specified I/O port or MMIO address.
1116 MMIO inter-register address stride is either 8-bit
1117 (mmio) or 32-bit (mmio32 or mmio32be).
1118 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1119 to be equivalent to 'mmio'. 'options' are specified
1120 in the same format described for "console=ttyS<n>"; if
1121 unspecified, the h/w is not initialized.
1125 Start an early, polled-mode console on a pl011 serial
1126 port at the specified address. The pl011 serial port
1127 must already be setup and configured. Options are not
1128 yet supported. If 'mmio32' is specified, then only
1129 the driver will use only 32-bit accessors to read/write
1130 the device registers.
1133 Start an early console on a litex serial port at the
1134 specified address. The serial port must already be
1135 setup and configured. Options are not yet supported.
1138 Start an early, polled-mode console on a meson serial
1139 port at the specified address. The serial port must
1140 already be setup and configured. Options are not yet
1144 Start an early, polled-mode console on an msm serial
1145 port at the specified address. The serial port
1146 must already be setup and configured. Options are not
1149 msm_serial_dm,<addr>
1150 Start an early, polled-mode console on an msm serial
1151 dm port at the specified address. The serial port
1152 must already be setup and configured. Options are not
1156 Start an early, polled-mode console on a serial port
1157 of an Actions Semi SoC, such as S500 or S900, at the
1158 specified address. The serial port must already be
1159 setup and configured. Options are not yet supported.
1162 Start an early, polled-mode console on a serial port
1163 of an RDA Micro SoC, such as RDA8810PL, at the
1164 specified address. The serial port must already be
1165 setup and configured. Options are not yet supported.
1168 Use RISC-V SBI (Supervisor Binary Interface) for early
1171 smh Use ARM semihosting calls for early console.
1179 Use early console provided by serial driver available
1180 on Samsung SoCs, requires selecting proper type and
1181 a correct base address of the selected UART port. The
1182 serial port must already be setup and configured.
1183 Options are not yet supported.
1186 Start an early, polled-mode console on a lantiq serial
1187 (lqasc) port at the specified address. The serial port
1188 must already be setup and configured. Options are not
1193 Use early console provided by Freescale LP UART driver
1194 found on Freescale Vybrid and QorIQ LS1021A processors.
1195 A valid base address must be provided, and the serial
1196 port must already be setup and configured.
1200 Start an early, polled-mode, output-only console on the
1201 Freescale i.MX UART at the specified address. The UART
1202 must already be setup and configured.
1205 Start an early, polled-mode console on the
1206 Armada 3700 serial port at the specified
1207 address. The serial port must already be setup
1208 and configured. Options are not yet supported.
1211 Start an early, polled-mode console on a Qualcomm
1212 Generic Interface (GENI) based serial port at the
1213 specified address. The serial port must already be
1214 setup and configured. Options are not yet supported.
1217 Start an early, unaccelerated console on the EFI
1218 memory mapped framebuffer (if available). On cache
1219 coherent non-x86 systems that use system memory for
1220 the framebuffer, pass the 'ram' option so that it is
1221 mapped with the correct attributes.
1224 Use early console provided by Freescale LINFlexD UART
1225 serial driver for NXP S32V234 SoCs. A valid base
1226 address must be provided, and the serial port must
1227 already be setup and configured.
1229 earlyprintk= [X86,SH,ARM,M68k,S390]
1233 earlyprintk=serial[,ttySn[,baudrate]]
1234 earlyprintk=serial[,0x...[,baudrate]]
1235 earlyprintk=ttySn[,baudrate]
1236 earlyprintk=dbgp[debugController#]
1237 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1238 earlyprintk=xdbc[xhciController#]
1240 earlyprintk is useful when the kernel crashes before
1241 the normal console is initialized. It is not enabled by
1242 default because it has some cosmetic problems.
1244 Append ",keep" to not disable it when the real console
1247 Only one of vga, efi, serial, or usb debug port can
1250 Currently only ttyS0 and ttyS1 may be specified by
1251 name. Other I/O ports may be explicitly specified
1252 on some architectures (x86 and arm at least) by
1253 replacing ttySn with an I/O port address, like this:
1254 earlyprintk=serial,0x1008,115200
1255 You can find the port for a given device in
1256 /proc/tty/driver/serial:
1257 2: uart:ST16650V2 port:00001008 irq:18 ...
1259 Interaction with the standard serial driver is not
1262 The VGA and EFI output is eventually overwritten by
1265 The xen option can only be used in Xen domains.
1267 The sclp output can only be used on s390.
1269 The optional "force" to "pciserial" enables use of a
1270 PCI device even when its classcode is not of the
1273 edac_report= [HW,EDAC] Control how to report EDAC event
1274 Format: {"on" | "off" | "force"}
1275 on: enable EDAC to report H/W event. May be overridden
1276 by other higher priority error reporting module.
1277 off: disable H/W event reporting through EDAC.
1278 force: enforce the use of EDAC to report H/W event.
1281 ekgdboc= [X86,KGDB] Allow early kernel console debugging
1284 This is designed to be used in conjunction with
1285 the boot argument: earlyprintk=vga
1287 This parameter works in place of the kgdboc parameter
1288 but can only be used if the backing tty is available
1289 very early in the boot process. For early debugging
1290 via a serial port see kgdboc_earlycon instead.
1293 Format: {"off" | "on" | "skip[mbr]"}
1296 Format: { "debug", "disable_early_pci_dma",
1297 "nochunk", "noruntime", "nosoftreserve",
1298 "novamap", "no_disable_early_pci_dma" }
1299 debug: enable misc debug output.
1300 disable_early_pci_dma: disable the busmaster bit on all
1301 PCI bridges while in the EFI boot stub.
1302 nochunk: disable reading files in "chunks" in the EFI
1303 boot stub, as chunking can cause problems with some
1304 firmware implementations.
1305 noruntime : disable EFI runtime services support
1306 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1307 attribute may cause the kernel to reserve the
1308 memory range for a memory mapping driver to
1309 claim. Specify efi=nosoftreserve to disable this
1310 reservation and treat the memory by its base type
1311 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1312 novamap: do not call SetVirtualAddressMap().
1313 no_disable_early_pci_dma: Leave the busmaster bit set
1314 on all PCI bridges while in the EFI boot stub
1316 efi_no_storage_paranoia [EFI; X86]
1317 Using this parameter you can use more than 50% of
1318 your efi variable storage. Use this parameter only if
1319 you are really sure that your UEFI does sane gc and
1320 fulfills the spec otherwise your board may brick.
1322 efi_fake_mem= nn[KMG]@ss[KMG]:aa[,nn[KMG]@ss[KMG]:aa,..] [EFI; X86]
1323 Add arbitrary attribute to specific memory range by
1324 updating original EFI memory map.
1325 Region of memory which aa attribute is added to is
1328 If efi_fake_mem=2G@4G:0x10000,2G@0x10a0000000:0x10000
1329 is specified, EFI_MEMORY_MORE_RELIABLE(0x10000)
1330 attribute is added to range 0x100000000-0x180000000 and
1331 0x10a0000000-0x1120000000.
1333 If efi_fake_mem=8G@9G:0x40000 is specified, the
1334 EFI_MEMORY_SP(0x40000) attribute is added to
1335 range 0x240000000-0x43fffffff.
1337 Using this parameter you can do debugging of EFI memmap
1338 related features. For example, you can do debugging of
1339 Address Range Mirroring feature even if your box
1340 doesn't support it, or mark specific memory as
1343 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1344 that is to be dynamically loaded by Linux. If there are
1345 multiple variables with the same name but with different
1346 vendor GUIDs, all of them will be loaded. See
1347 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1350 eisa_irq_edge= [PARISC,HW]
1351 See header of drivers/parisc/eisa.c.
1354 See comment before function elanfreq_setup() in
1355 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1357 elfcorehdr=[size[KMG]@]offset[KMG] [IA64,PPC,SH,X86,S390]
1358 Specifies physical address of start of kernel core
1359 image elf header and optionally the size. Generally
1360 kexec loader will pass this option to capture kernel.
1361 See Documentation/admin-guide/kdump/kdump.rst for details.
1363 enable_mtrr_cleanup [X86]
1364 The kernel tries to adjust MTRR layout from continuous
1365 to discrete, to make X server driver able to add WB
1366 entry later. This parameter enables that.
1368 enable_timer_pin_1 [X86]
1369 Enable PIN 1 of APIC timer
1370 Can be useful to work around chipset bugs
1371 (in particular on some ATI chipsets).
1372 The kernel tries to set a reasonable default.
1374 enforcing [SELINUX] Set initial enforcing status.
1376 See security/selinux/Kconfig help text.
1377 0 -- permissive (log only, no denials).
1378 1 -- enforcing (deny and log).
1380 Value can be changed at runtime via
1381 /sys/fs/selinux/enforce.
1384 Disable Error Record Serialization Table (ERST)
1387 ether= [HW,NET] Ethernet cards parameters
1388 This option is obsoleted by the "netdev=" option, which
1389 has equivalent usage. See its documentation for details.
1393 Permit 'security.evm' to be updated regardless of
1394 current integrity status.
1399 fail_make_request=[KNL]
1400 General fault injection mechanism.
1401 Format: <interval>,<probability>,<space>,<times>
1402 See also Documentation/fault-injection/.
1405 Format: { initns | none }
1406 See Documentation/admin-guide/sysctl/net.rst for
1407 fb_tunnels_only_for_init_ns
1410 See Documentation/admin-guide/blockdev/floppy.rst.
1412 force_pal_cache_flush
1413 [IA-64] Avoid check_sal_cache_flush which may hang on
1414 buggy SAL_CACHE_FLUSH implementations. Using this
1415 parameter will force ia64_sal_cache_flush to call
1416 ia64_pal_cache_flush instead of SAL_CACHE_FLUSH.
1419 Forcefully enable Physical Address Extension (PAE).
1420 Many Pentium M systems disable PAE but may have a
1421 functionally usable PAE implementation.
1422 Warning: use of this parameter will taint the kernel
1423 and may cause unknown problems.
1426 [FTRACE] will set and start the specified tracer
1427 as early as possible in order to facilitate early
1430 ftrace_dump_on_oops[=orig_cpu]
1431 [FTRACE] will dump the trace buffers on oops.
1432 If no parameter is passed, ftrace will dump
1433 buffers of all CPUs, but if you pass orig_cpu, it will
1434 dump only the buffer of the CPU that triggered the
1437 ftrace_filter=[function-list]
1438 [FTRACE] Limit the functions traced by the function
1439 tracer at boot up. function-list is a comma-separated
1440 list of functions. This list can be changed at run
1441 time by the set_ftrace_filter file in the debugfs
1444 ftrace_notrace=[function-list]
1445 [FTRACE] Do not trace the functions specified in
1446 function-list. This list can be changed at run time
1447 by the set_ftrace_notrace file in the debugfs
1450 ftrace_graph_filter=[function-list]
1451 [FTRACE] Limit the top level callers functions traced
1452 by the function graph tracer at boot up.
1453 function-list is a comma-separated list of functions
1454 that can be changed at run time by the
1455 set_graph_function file in the debugfs tracing directory.
1457 ftrace_graph_notrace=[function-list]
1458 [FTRACE] Do not trace from the functions specified in
1459 function-list. This list is a comma-separated list of
1460 functions that can be changed at run time by the
1461 set_graph_notrace file in the debugfs tracing directory.
1463 ftrace_graph_max_depth=<uint>
1464 [FTRACE] Used with the function graph tracer. This is
1465 the max depth it will trace into a function. This value
1466 can be changed at run time by the max_graph_depth file
1467 in the tracefs tracing directory. default: 0 (no limit)
1469 fw_devlink= [KNL] Create device links between consumer and supplier
1470 devices by scanning the firmware to infer the
1471 consumer/supplier relationships. This feature is
1472 especially useful when drivers are loaded as modules as
1473 it ensures proper ordering of tasks like device probing
1474 (suppliers first, then consumers), supplier boot state
1475 clean up (only after all consumers have probed),
1476 suspend/resume & runtime PM (consumers first, then
1478 Format: { off | permissive | on | rpm }
1479 off -- Don't create device links from firmware info.
1480 permissive -- Create device links from firmware info
1481 but use it only for ordering boot state clean
1482 up (sync_state() calls).
1483 on -- Create device links from firmware info and use it
1484 to enforce probe and suspend/resume ordering.
1485 rpm -- Like "on", but also use to order runtime PM.
1487 fw_devlink.strict=<bool>
1488 [KNL] Treat all inferred dependencies as mandatory
1489 dependencies. This only applies for fw_devlink=on|rpm.
1493 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1494 support via parallel port (up to 5 devices per port)
1495 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1496 See also Documentation/input/devices/joystick-parport.rst
1500 gart_fix_e820= [X86-64] disable the fix e820 for K8 GART
1504 gather_data_sampling=
1505 [X86,INTEL] Control the Gather Data Sampling (GDS)
1508 Gather Data Sampling is a hardware vulnerability which
1509 allows unprivileged speculative access to data which was
1510 previously stored in vector registers.
1512 This issue is mitigated by default in updated microcode.
1513 The mitigation may have a performance impact but can be
1514 disabled. On systems without the microcode mitigation
1515 disabling AVX serves as a mitigation.
1517 force: Disable AVX to mitigate systems without
1518 microcode mitigation. No effect if the microcode
1519 mitigation is present. Known to cause crashes in
1520 userspace with buggy AVX enumeration.
1522 off: Disable GDS mitigation.
1524 gcov_persist= [GCOV] When non-zero (default), profiling data for
1525 kernel modules is saved and remains accessible via
1526 debugfs, even when the module is unloaded/reloaded.
1527 When zero, profiling data is discarded and associated
1528 debugfs files are removed at module unload time.
1530 goldfish [X86] Enable the goldfish android emulator platform.
1531 Don't use this when you are not running on the
1534 gpio-mockup.gpio_mockup_ranges
1535 [HW] Sets the ranges of gpiochip of for this device.
1536 Format: <start1>,<end1>,<start2>,<end2>...
1537 gpio-mockup.gpio_mockup_named_lines
1538 [HW] Let the driver know GPIO lines should be named.
1540 gpt [EFI] Forces disk with valid GPT signature but
1541 invalid Protective MBR to be treated as GPT. If the
1542 primary GPT is corrupted, it enables the backup/alternate
1543 GPT to be used instead.
1545 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1546 the "Enable 0" bit of the configuration register.
1549 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1550 the "Enable 0" bit of the configuration register.
1553 grcan.select= [HW] Select which physical interface to use.
1556 grcan.txsize= [HW] Sets the size of the tx buffer.
1557 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1559 grcan.rxsize= [HW] Sets the size of the rx buffer.
1560 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1563 hardlockup_all_cpu_backtrace=
1564 [KNL] Should the hard-lockup detector generate
1565 backtraces on all cpus.
1568 hashdist= [KNL,NUMA] Large hashes allocated during boot
1569 are distributed across NUMA nodes. Defaults on
1570 for 64-bit NUMA, off otherwise.
1571 Format: 0 | 1 (for off | on)
1573 hcl= [IA-64] SGI's Hardware Graph compatibility layer
1575 hd= [EIDE] (E)IDE hard drive subsystem geometry
1576 Format: <cyl>,<head>,<sect>
1579 Disable Hardware Error Source Table (HEST) support;
1580 corresponding firmware-first mode error processing
1581 logic will be disabled.
1583 highmem=nn[KMG] [KNL,BOOT] forces the highmem zone to have an exact
1584 size of <nn>. This works even on boxes that have no
1585 highmem otherwise. This also works to reduce highmem
1586 size on bigger boxes.
1588 highres= [KNL] Enable/disable high resolution timer mode.
1589 Valid parameters: "on", "off"
1594 hpet= [X86-32,HPET] option to control HPET usage
1595 Format: { enable (default) | disable | force |
1597 disable: disable HPET and use PIT instead
1598 force: allow force enabled of undocumented chips (ICH4,
1600 verbose: show contents of HPET registers during setup
1602 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1603 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1605 hugetlb_cma= [HW,CMA] The size of a CMA area used for allocation
1606 of gigantic hugepages.
1609 Reserve a CMA area of given size and allocate gigantic
1610 hugepages using the CMA allocator. If enabled, the
1611 boot-time allocation of gigantic hugepages is skipped.
1613 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1614 If this follows hugepagesz (below), it specifies
1615 the number of pages of hugepagesz to be allocated.
1616 If this is the first HugeTLB parameter on the command
1617 line, it specifies the number of pages to allocate for
1618 the default huge page size. See also
1619 Documentation/admin-guide/mm/hugetlbpage.rst.
1623 [HW] The size of the HugeTLB pages. This is used in
1624 conjunction with hugepages (above) to allocate huge
1625 pages of a specific size at boot. The pair
1626 hugepagesz=X hugepages=Y can be specified once for
1627 each supported huge page size. Huge page sizes are
1628 architecture dependent. See also
1629 Documentation/admin-guide/mm/hugetlbpage.rst.
1632 hugetlb_free_vmemmap=
1633 [KNL] Reguires CONFIG_HUGETLB_PAGE_FREE_VMEMMAP
1635 Allows heavy hugetlb users to free up some more
1636 memory (6 * PAGE_SIZE for each 2MB hugetlb page).
1637 Format: { on | off (default) }
1639 on: enable the feature
1640 off: disable the feature
1642 Built with CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON=y,
1645 This is not compatible with memory_hotplug.memmap_on_memory.
1646 If both parameters are enabled, hugetlb_free_vmemmap takes
1647 precedence over memory_hotplug.memmap_on_memory.
1650 [KNL] Should the hung task detector generate panics.
1653 A value of 1 instructs the kernel to panic when a
1654 hung task is detected. The default value is controlled
1655 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1656 option. The value selected by this boot parameter can
1657 be changed later by the kernel.hung_task_panic sysctl.
1659 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1660 terminal devices. Valid values: 0..8
1661 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1662 If specified, z/VM IUCV HVC accepts connections
1663 from listed z/VM user IDs only.
1665 hv_nopvspin [X86,HYPER_V] Disables the paravirt spinlock optimizations
1666 which allow the hypervisor to 'idle' the
1667 guest on lock contention.
1670 Do not unregister boot console at start. This is only
1671 useful for debugging when something happens in the window
1672 between unregistering the boot console and initializing
1675 i2c_bus= [HW] Override the default board specific I2C bus speed
1676 or register an additional I2C bus that is not
1677 registered from board initialization code.
1681 i8042.debug [HW] Toggle i8042 debug mode
1682 i8042.unmask_kbd_data
1683 [HW] Enable printing of interrupt data from the KBD port
1684 (disabled by default, and as a pre-condition
1685 requires that i8042.debug=1 be enabled)
1686 i8042.direct [HW] Put keyboard port into non-translated mode
1687 i8042.dumbkbd [HW] Pretend that controller can only read data from
1688 keyboard and cannot control its state
1689 (Don't attempt to blink the leds)
1690 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1691 i8042.nokbd [HW] Don't check/create keyboard port
1692 i8042.noloop [HW] Disable the AUX Loopback command while probing
1694 i8042.nomux [HW] Don't check presence of an active multiplexing
1696 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1698 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1699 i8042.reset [HW] Reset the controller during init, cleanup and
1700 suspend-to-ram transitions, only during s2r
1701 transitions, or never reset
1702 Format: { 1 | Y | y | 0 | N | n }
1703 1, Y, y: always reset controller
1704 0, N, n: don't ever reset controller
1705 Default: only on s2r transitions on x86; most other
1706 architectures force reset to be always executed
1707 i8042.unlock [HW] Unlock (ignore) the keylock
1708 i8042.kbdreset [HW] Reset device connected to KBD port
1710 [HW] Allow deferred probing upon i8042 probe errors
1714 i8k.ignore_dmi [HW] Continue probing hardware even if DMI data
1715 indicates that the driver is running on unsupported
1717 i8k.force [HW] Activate i8k driver even if SMM BIOS signature
1718 does not match list of supported models.
1720 [HW] Report power status in /proc/i8k
1721 (disabled by default)
1722 i8k.restricted [HW] Allow controlling fans only if SYS_ADMIN
1725 i915.invert_brightness=
1726 [DRM] Invert the sense of the variable that is used to
1727 set the brightness of the panel backlight. Normally a
1728 brightness value of 0 indicates backlight switched off,
1729 and the maximum of the brightness value sets the backlight
1730 to maximum brightness. If this parameter is set to 0
1731 (default) and the machine requires it, or this parameter
1732 is set to 1, a brightness value of 0 sets the backlight
1733 to maximum brightness, and the maximum of the brightness
1734 value switches the backlight off.
1735 -1 -- never invert brightness
1736 0 -- machine default
1737 1 -- force brightness inversion
1740 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1742 ide-core.nodma= [HW] (E)IDE subsystem
1743 Format: =0.0 to prevent dma on hda, =0.1 hdb =1.0 hdc
1744 .vlb_clock .pci_clock .noflush .nohpa .noprobe .nowerr
1745 .cdrom .chs .ignore_cable are additional options
1746 See Documentation/ide/ide.rst.
1748 ide-generic.probe-mask= [HW] (E)IDE subsystem
1750 Probe mask for legacy ISA IDE ports. Depending on
1751 platform up to 6 ports are supported, enabled by
1752 setting corresponding bits in the mask to 1. The
1753 default value is 0x0, which has a special meaning.
1754 On systems that have PCI, it triggers scanning the
1755 PCI bus for the first and the second port, which
1756 are then probed. On systems without PCI the value
1757 of 0x0 enables probing the two first ports as if it
1760 ide-pci-generic.all-generic-ide [HW] (E)IDE subsystem
1761 Claim all unknown PCI IDE storage controllers.
1764 Format: idle=poll, idle=halt, idle=nomwait
1765 Poll forces a polling idle loop that can slightly
1766 improve the performance of waking up a idle CPU, but
1767 will use a lot of power and make the system run hot.
1769 idle=halt: Halt is forced to be used for CPU idle.
1770 In such case C2/C3 won't be used again.
1771 idle=nomwait: Disable mwait for CPU C-states
1775 Allow force disabling of Shared Virtual Memory (SVA)
1776 support for the idxd driver. By default it is set to
1779 idxd.tc_override= [HW]
1781 Allow override of default traffic class configuration
1782 for the device. By default it is set to false (0).
1784 ieee754= [MIPS] Select IEEE Std 754 conformance mode
1785 Format: { strict | legacy | 2008 | relaxed }
1788 Choose which programs will be accepted for execution
1789 based on the IEEE 754 NaN encoding(s) supported by
1790 the FPU and the NaN encoding requested with the value
1791 of an ELF file header flag individually set by each
1792 binary. Hardware implementations are permitted to
1793 support either or both of the legacy and the 2008 NaN
1796 Available settings are as follows:
1797 strict accept binaries that request a NaN encoding
1798 supported by the FPU
1799 legacy only accept legacy-NaN binaries, if supported
1801 2008 only accept 2008-NaN binaries, if supported
1803 relaxed accept any binaries regardless of whether
1804 supported by the FPU
1806 The FPU emulator is always able to support both NaN
1807 encodings, so if no FPU hardware is present or it has
1808 been disabled with 'nofpu', then the settings of
1809 'legacy' and '2008' strap the emulator accordingly,
1810 'relaxed' straps the emulator for both legacy-NaN and
1811 2008-NaN, whereas 'strict' enables legacy-NaN only on
1812 legacy processors and both NaN encodings on MIPS32 or
1815 The setting for ABS.fmt/NEG.fmt instruction execution
1816 mode generally follows that for the NaN encoding,
1817 except where unsupported by hardware.
1819 ignore_loglevel [KNL]
1820 Ignore loglevel setting - this will print /all/
1821 kernel messages to the console. Useful for debugging.
1822 We also add it as printk module parameter, so users
1823 could change it dynamically, usually by
1824 /sys/module/printk/parameters/ignore_loglevel.
1827 Ignore RLIMIT_DATA setting for data mappings,
1828 print warning at first misuse. Can be changed via
1829 /sys/module/kernel/parameters/ignore_rlimit_data.
1831 ihash_entries= [KNL]
1832 Set number of hash buckets for inode cache.
1834 ima_appraise= [IMA] appraise integrity measurements
1835 Format: { "off" | "enforce" | "fix" | "log" }
1838 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
1839 The builtin appraise policy appraises all files
1842 ima_canonical_fmt [IMA]
1843 Use the canonical format for the binary runtime
1844 measurements, instead of host native format.
1847 Format: { md5 | sha1 | rmd160 | sha256 | sha384
1851 The list of supported hash algorithms is defined
1852 in crypto/hash_info.h.
1855 The builtin policies to load during IMA setup.
1856 Format: "tcb | appraise_tcb | secure_boot |
1857 fail_securely | critical_data"
1859 The "tcb" policy measures all programs exec'd, files
1860 mmap'd for exec, and all files opened with the read
1861 mode bit set by either the effective uid (euid=0) or
1864 The "appraise_tcb" policy appraises the integrity of
1865 all files owned by root.
1867 The "secure_boot" policy appraises the integrity
1868 of files (eg. kexec kernel image, kernel modules,
1869 firmware, policy, etc) based on file signatures.
1871 The "fail_securely" policy forces file signature
1872 verification failure also on privileged mounted
1873 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
1876 The "critical_data" policy measures kernel integrity
1879 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
1880 Load a policy which meets the needs of the Trusted
1881 Computing Base. This means IMA will measure all
1882 programs exec'd, files mmap'd for exec, and all files
1883 opened for read by uid=0.
1886 Select one of defined IMA measurements template formats.
1887 Formats: { "ima" | "ima-ng" | "ima-sig" }
1891 [IMA] Define a custom template format.
1892 Format: { "field1|...|fieldN" }
1894 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
1895 Format: <min_file_size>
1896 Set the minimal file size for using asynchronous hash.
1897 If left unspecified, ahash usage is disabled.
1899 ahash performance varies for different data sizes on
1900 different crypto accelerators. This option can be used
1901 to achieve the best performance for a particular HW.
1903 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
1905 Set hashing buffer size. Default: 4k.
1907 ahash performance varies for different chunk sizes on
1908 different crypto accelerators. This option can be used
1909 to achieve best performance for particular HW.
1913 Run specified binary instead of /sbin/init as init
1916 initcall_debug [KNL] Trace initcalls as they are executed. Useful
1917 for working out where the kernel is dying during
1920 initcall_blacklist= [KNL] Do not execute a comma-separated list of
1921 initcall functions. Useful for debugging built-in
1922 modules and initcalls.
1924 initramfs_async= [KNL]
1927 This parameter controls whether the initramfs
1928 image is unpacked asynchronously, concurrently
1929 with devices being probed and
1930 initialized. This should normally just work,
1931 but as a debugging aid, one can get the
1932 historical behaviour of the initramfs
1933 unpacking being completed before device_ and
1936 initrd= [BOOT] Specify the location of the initial ramdisk
1938 initrdmem= [KNL] Specify a physical address and size from which to
1939 load the initrd. If an initrd is compiled in or
1940 specified in the bootparams, it takes priority over this
1942 Format: ss[KMG],nn[KMG]
1945 init_on_alloc= [MM] Fill newly allocated pages and heap objects with
1948 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
1950 init_on_free= [MM] Fill freed pages and heap objects with zeroes.
1952 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
1954 init_pkru= [X86] Specify the default memory protection keys rights
1955 register contents for all processes. 0x55555554 by
1956 default (disallow access to all but pkey 0). Can
1957 override in debugfs after boot.
1959 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
1962 int_pln_enable [X86] Enable power limit notification interrupt
1964 integrity_audit=[IMA]
1965 Format: { "0" | "1" }
1966 0 -- basic integrity auditing messages. (Default)
1967 1 -- additional integrity auditing messages.
1969 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
1971 Enable intel iommu driver.
1973 Disable intel iommu driver.
1974 igfx_off [Default Off]
1975 By default, gfx is mapped as normal device. If a gfx
1976 device has a dedicated DMAR unit, the DMAR unit is
1977 bypassed by not enabling DMAR with this option. In
1978 this case, gfx device will use physical address for
1980 strict [Default Off]
1981 Deprecated, equivalent to iommu.strict=1.
1982 sp_off [Default Off]
1983 By default, super page will be supported if Intel IOMMU
1984 has the capability. With this option, super page will
1987 Enable the Intel IOMMU scalable mode if the hardware
1988 advertises that it has support for the scalable mode
1991 Disallow use of the Intel IOMMU scalable mode.
1992 tboot_noforce [Default Off]
1993 Do not force the Intel IOMMU enabled under tboot.
1994 By default, tboot will force Intel IOMMU on, which
1995 could harm performance of some high-throughput
1996 devices like 40GBit network cards, even if identity
1998 Note that using this option lowers the security
1999 provided by tboot because it makes the system
2000 vulnerable to DMA attacks.
2002 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2003 0 disables intel_idle and fall back on acpi_idle.
2004 1 to 9 specify maximum depth of C-state.
2008 Do not enable intel_pstate as the default
2009 scaling driver for the supported processors
2011 Use intel_pstate as a scaling driver, but configure it
2012 to work with generic cpufreq governors (instead of
2013 enabling its internal governor). This mode cannot be
2014 used along with the hardware-managed P-states (HWP)
2017 Enable intel_pstate on systems that prohibit it by default
2018 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2019 instead of acpi-cpufreq may disable platform features, such
2020 as thermal controls and power capping, that rely on ACPI
2021 P-States information being indicated to OSPM and therefore
2022 should be used with caution. This option does not work with
2023 processors that aren't supported by the intel_pstate driver
2024 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2026 Do not enable hardware P state control (HWP)
2029 Only load intel_pstate on systems which support
2030 hardware P state control (HWP) if available.
2032 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2033 Description Table, specifies preferred power management
2034 profile as "Enterprise Server" or "Performance Server",
2035 then this feature is turned on by default.
2037 Allow per-logical-CPU P-State performance control limits using
2038 cpufreq sysfs interface
2040 intremap= [X86-64, Intel-IOMMU]
2041 on enable Interrupt Remapping (default)
2042 off disable Interrupt Remapping
2043 nosid disable Source ID checking
2045 BIOS x2APIC opt-out request will be ignored
2046 nopost disable Interrupt Posting
2048 iomem= Disable strict checking of access to MMIO memory
2049 strict regions from userspace.
2064 nobypass [PPC/POWERNV]
2065 Disable IOMMU bypass, using IOMMU for PCI devices.
2067 iommu.forcedac= [ARM64, X86] Control IOVA allocation for PCI devices.
2068 Format: { "0" | "1" }
2069 0 - Try to allocate a 32-bit DMA address first, before
2070 falling back to the full range if needed.
2071 1 - Allocate directly from the full usable range,
2072 forcing Dual Address Cycle for PCI cards supporting
2073 greater than 32-bit addressing.
2075 iommu.strict= [ARM64, X86] Configure TLB invalidation behaviour
2076 Format: { "0" | "1" }
2078 Request that DMA unmap operations use deferred
2079 invalidation of hardware TLBs, for increased
2080 throughput at the cost of reduced device isolation.
2081 Will fall back to strict mode if not supported by
2082 the relevant IOMMU driver.
2084 DMA unmap operations invalidate IOMMU hardware TLBs
2086 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2087 Note: on x86, strict mode specified via one of the
2088 legacy driver-specific options takes precedence.
2091 [ARM64, X86] Configure DMA to bypass the IOMMU by default.
2092 Format: { "0" | "1" }
2093 0 - Use IOMMU translation for DMA.
2094 1 - Bypass the IOMMU for DMA.
2095 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2097 io7= [HW] IO7 for Marvel-based Alpha systems
2098 See comment before marvel_specify_io7 in
2099 arch/alpha/kernel/core_marvel.c.
2101 io_delay= [X86] I/O delay method
2103 Standard port 0x80 based delay
2105 Alternate port 0xed based delay (needed on some systems)
2107 Simple two microseconds delay
2112 See Documentation/admin-guide/nfs/nfsroot.rst.
2114 ipcmni_extend [KNL] Extend the maximum number of unique System V
2115 IPC identifiers from 32,768 to 16,777,216.
2117 irqaffinity= [SMP] Set the default irq affinity mask
2118 The argument is a cpu list, as described above.
2120 irqchip.gicv2_force_probe=
2123 Force the kernel to look for the second 4kB page
2124 of a GICv2 controller even if the memory range
2125 exposed by the device tree is too small.
2127 irqchip.gicv3_nolpi=
2129 Force the kernel to ignore the availability of
2130 LPIs (and by consequence ITSs). Intended for system
2131 that use the kernel as a bootloader, and thus want
2132 to let secondary kernels in charge of setting up
2135 irqchip.gicv3_pseudo_nmi= [ARM64]
2136 Enables support for pseudo-NMIs in the kernel. This
2137 requires the kernel to be built with
2138 CONFIG_ARM64_PSEUDO_NMI.
2141 When an interrupt is not handled search all handlers
2142 for it. Intended to get systems with badly broken
2146 When an interrupt is not handled search all handlers
2147 for it. Also check all handlers each timer
2148 interrupt. Intended to get systems with badly broken
2152 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2154 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2155 [Deprecated - use cpusets instead]
2156 Format: [flag-list,]<cpu-list>
2158 Specify one or more CPUs to isolate from disturbances
2159 specified in the flag list (default: domain):
2162 Disable the tick when a single task runs.
2164 A residual 1Hz tick is offloaded to workqueues, which you
2165 need to affine to housekeeping through the global
2166 workqueue's affinity configured via the
2167 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2168 by using the 'domain' flag described below.
2170 NOTE: by default the global workqueue runs on all CPUs,
2171 so to protect individual CPUs the 'cpumask' file has to
2172 be configured manually after bootup.
2175 Isolate from the general SMP balancing and scheduling
2176 algorithms. Note that performing domain isolation this way
2177 is irreversible: it's not possible to bring back a CPU to
2178 the domains once isolated through isolcpus. It's strongly
2179 advised to use cpusets instead to disable scheduler load
2180 balancing through the "cpuset.sched_load_balance" file.
2181 It offers a much more flexible interface where CPUs can
2182 move in and out of an isolated set anytime.
2184 You can move a process onto or off an "isolated" CPU via
2185 the CPU affinity syscalls or cpuset.
2186 <cpu number> begins at 0 and the maximum value is
2187 "number of CPUs in system - 1".
2191 Isolate from being targeted by managed interrupts
2192 which have an interrupt mask containing isolated
2193 CPUs. The affinity of managed interrupts is
2194 handled by the kernel and cannot be changed via
2195 the /proc/irq/* interfaces.
2197 This isolation is best effort and only effective
2198 if the automatically assigned interrupt mask of a
2199 device queue contains isolated and housekeeping
2200 CPUs. If housekeeping CPUs are online then such
2201 interrupts are directed to the housekeeping CPU
2202 so that IO submitted on the housekeeping CPU
2203 cannot disturb the isolated CPU.
2205 If a queue's affinity mask contains only isolated
2206 CPUs then this parameter has no effect on the
2207 interrupt routing decision, though interrupts are
2208 only delivered when tasks running on those
2209 isolated CPUs submit IO. IO submitted on
2210 housekeeping CPUs has no influence on those
2213 The format of <cpu-list> is described above.
2217 ivrs_ioapic [HW,X86-64]
2218 Provide an override to the IOAPIC-ID<->DEVICE-ID
2219 mapping provided in the IVRS ACPI table.
2220 By default, PCI segment is 0, and can be omitted.
2222 For example, to map IOAPIC-ID decimal 10 to
2223 PCI segment 0x1 and PCI device 00:14.0,
2224 write the parameter as:
2225 ivrs_ioapic=10@0001:00:14.0
2228 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2229 write the parameter as:
2230 ivrs_ioapic[10]=00:14.0
2231 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2232 PCI device 00:14.0 write the parameter as:
2233 ivrs_ioapic[10]=0001:00:14.0
2235 ivrs_hpet [HW,X86-64]
2236 Provide an override to the HPET-ID<->DEVICE-ID
2237 mapping provided in the IVRS ACPI table.
2238 By default, PCI segment is 0, and can be omitted.
2240 For example, to map HPET-ID decimal 10 to
2241 PCI segment 0x1 and PCI device 00:14.0,
2242 write the parameter as:
2243 ivrs_hpet=10@0001:00:14.0
2246 * To map HPET-ID decimal 0 to PCI device 00:14.0
2247 write the parameter as:
2248 ivrs_hpet[0]=00:14.0
2249 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2250 PCI device 00:14.0 write the parameter as:
2251 ivrs_ioapic[10]=0001:00:14.0
2253 ivrs_acpihid [HW,X86-64]
2254 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2255 mapping provided in the IVRS ACPI table.
2256 By default, PCI segment is 0, and can be omitted.
2258 For example, to map UART-HID:UID AMD0020:0 to
2259 PCI segment 0x1 and PCI device ID 00:14.5,
2260 write the parameter as:
2261 ivrs_acpihid=AMD0020:0@0001:00:14.5
2264 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2265 PCI device ID 00:14.5, write the parameter as:
2266 ivrs_acpihid[00:14.5]=AMD0020:0
2267 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2268 PCI device ID 00:14.5, write the parameter as:
2269 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2271 js= [HW,JOY] Analog joystick
2272 See Documentation/input/joydev/joystick.rst.
2275 When CONFIG_RANDOMIZE_BASE is set, this disables
2276 kernel and module base offset ASLR (Address Space
2277 Layout Randomization).
2280 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2281 report on every invalid memory access. Without this
2282 parameter KASAN will print report only for the first
2287 kernelcore= [KNL,X86,IA-64,PPC]
2288 Format: nn[KMGTPE] | nn% | "mirror"
2289 This parameter specifies the amount of memory usable by
2290 the kernel for non-movable allocations. The requested
2291 amount is spread evenly throughout all nodes in the
2292 system as ZONE_NORMAL. The remaining memory is used for
2293 movable memory in its own zone, ZONE_MOVABLE. In the
2294 event, a node is too small to have both ZONE_NORMAL and
2295 ZONE_MOVABLE, kernelcore memory will take priority and
2296 other nodes will have a larger ZONE_MOVABLE.
2298 ZONE_MOVABLE is used for the allocation of pages that
2299 may be reclaimed or moved by the page migration
2300 subsystem. Note that allocations like PTEs-from-HighMem
2301 still use the HighMem zone if it exists, and the Normal
2302 zone if it does not.
2304 It is possible to specify the exact amount of memory in
2305 the form of "nn[KMGTPE]", a percentage of total system
2306 memory in the form of "nn%", or "mirror". If "mirror"
2307 option is specified, mirrored (reliable) memory is used
2308 for non-movable allocations and remaining memory is used
2309 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2310 are exclusive, so you cannot specify multiple forms.
2312 kgdbdbgp= [KGDB,HW] kgdb over EHCI usb debug port.
2313 Format: <Controller#>[,poll interval]
2314 The controller # is the number of the ehci usb debug
2315 port as it is probed via PCI. The poll interval is
2316 optional and is the number seconds in between
2317 each poll cycle to the debug port in case you need
2318 the functionality for interrupting the kernel with
2319 gdb or control-c on the dbgp connection. When
2320 not using this parameter you use sysrq-g to break into
2321 the kernel debugger.
2323 kgdboc= [KGDB,HW] kgdb over consoles.
2324 Requires a tty driver that supports console polling,
2325 or a supported polling keyboard driver (non-usb).
2326 Serial only format: <serial_device>[,baud]
2327 keyboard only format: kbd
2328 keyboard and serial format: kbd,<serial_device>[,baud]
2329 Optional Kernel mode setting:
2330 kms, kbd format: kms,kbd
2331 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2333 kgdboc_earlycon= [KGDB,HW]
2334 If the boot console provides the ability to read
2335 characters and can work in polling mode, you can use
2336 this parameter to tell kgdb to use it as a backend
2337 until the normal console is registered. Intended to
2338 be used together with the kgdboc parameter which
2339 specifies the normal console to transition to.
2341 The name of the early console should be specified
2342 as the value of this parameter. Note that the name of
2343 the early console might be different than the tty
2344 name passed to kgdboc. It's OK to leave the value
2345 blank and the first boot console that implements
2346 read() will be picked.
2348 kgdbwait [KGDB] Stop kernel execution and enter the
2349 kernel debugger at the earliest opportunity.
2351 kmac= [MIPS] Korina ethernet MAC address.
2352 Configure the RouterBoard 532 series on-chip
2353 Ethernet adapter MAC address.
2355 kmemleak= [KNL] Boot-time kmemleak enable/disable
2356 Valid arguments: on, off
2358 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2361 kprobe_event=[probe-list]
2362 [FTRACE] Add kprobe events and enable at boot time.
2363 The probe-list is a semicolon delimited list of probe
2364 definitions. Each definition is same as kprobe_events
2365 interface, but the parameters are comma delimited.
2366 For example, to add a kprobe event on vfs_read with
2367 arg1 and arg2, add to the command line;
2369 kprobe_event=p,vfs_read,$arg1,$arg2
2371 See also Documentation/trace/kprobetrace.rst "Kernel
2372 Boot Parameter" section.
2374 kpti= [ARM64] Control page table isolation of user
2375 and kernel address spaces.
2376 Default: enabled on cores which need mitigation.
2380 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2381 Default is 0 (don't ignore, but inject #GP)
2383 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2384 Default is false (don't support).
2386 kvm.mmu_audit= [KVM] This is a R/W parameter which allows audit
2391 [KVM] Controls the software workaround for the
2392 X86_BUG_ITLB_MULTIHIT bug.
2393 force : Always deploy workaround.
2394 off : Never deploy workaround.
2395 auto : Deploy workaround based on the presence of
2396 X86_BUG_ITLB_MULTIHIT.
2400 If the software workaround is enabled for the host,
2401 guests do need not to enable it for nested guests.
2403 kvm.nx_huge_pages_recovery_ratio=
2404 [KVM] Controls how many 4KiB pages are periodically zapped
2405 back to huge pages. 0 disables the recovery, otherwise if
2406 the value is N KVM will zap 1/Nth of the 4KiB pages every
2407 minute. The default is 60.
2409 kvm-amd.nested= [KVM,AMD] Allow nested virtualization in KVM/SVM.
2410 Default is 1 (enabled)
2412 kvm-amd.npt= [KVM,AMD] Disable nested paging (virtualized MMU)
2414 Default is 1 (enabled) if in 64-bit or 32-bit PAE mode.
2417 [KVM,ARM] Select one of KVM/arm64's modes of operation.
2419 nvhe: Standard nVHE-based mode, without support for
2422 protected: nVHE-based mode with support for guests whose
2423 state is kept private from the host.
2424 Not valid if the kernel is running in EL2.
2426 Defaults to VHE/nVHE based on hardware support.
2428 kvm-arm.vgic_v3_group0_trap=
2429 [KVM,ARM] Trap guest accesses to GICv3 group-0
2432 kvm-arm.vgic_v3_group1_trap=
2433 [KVM,ARM] Trap guest accesses to GICv3 group-1
2436 kvm-arm.vgic_v3_common_trap=
2437 [KVM,ARM] Trap guest accesses to GICv3 common
2440 kvm-arm.vgic_v4_enable=
2441 [KVM,ARM] Allow use of GICv4 for direct injection of
2444 kvm_cma_resv_ratio=n [PPC]
2445 Reserves given percentage from system memory area for
2446 contiguous memory allocation for KVM hash pagetable
2448 By default it reserves 5% of total system memory.
2452 kvm-intel.ept= [KVM,Intel] Disable extended page tables
2453 (virtualized MMU) support on capable Intel chips.
2454 Default is 1 (enabled)
2456 kvm-intel.emulate_invalid_guest_state=
2457 [KVM,Intel] Disable emulation of invalid guest state.
2458 Ignored if kvm-intel.enable_unrestricted_guest=1, as
2459 guest state is never invalid for unrestricted guests.
2460 This param doesn't apply to nested guests (L2), as KVM
2461 never emulates invalid L2 guest state.
2462 Default is 1 (enabled)
2464 kvm-intel.flexpriority=
2465 [KVM,Intel] Disable FlexPriority feature (TPR shadow).
2466 Default is 1 (enabled)
2469 [KVM,Intel] Enable VMX nesting (nVMX).
2470 Default is 0 (disabled)
2472 kvm-intel.unrestricted_guest=
2473 [KVM,Intel] Disable unrestricted guest feature
2474 (virtualized real and unpaged mode) on capable
2475 Intel chips. Default is 1 (enabled)
2477 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2480 Valid arguments: never, cond, always
2482 always: L1D cache flush on every VMENTER.
2483 cond: Flush L1D on VMENTER only when the code between
2484 VMEXIT and VMENTER can leak host memory.
2485 never: Disables the mitigation
2487 Default is cond (do L1 cache flush in specific instances)
2489 kvm-intel.vpid= [KVM,Intel] Disable Virtual Processor Identification
2490 feature (tagged TLBs) on capable Intel chips.
2491 Default is 1 (enabled)
2493 l1d_flush= [X86,INTEL]
2494 Control mitigation for L1D based snooping vulnerability.
2496 Certain CPUs are vulnerable to an exploit against CPU
2497 internal buffers which can forward information to a
2498 disclosure gadget under certain conditions.
2500 In vulnerable processors, the speculatively
2501 forwarded data can be used in a cache side channel
2502 attack, to access data to which the attacker does
2503 not have direct access.
2505 This parameter controls the mitigation. The
2508 on - enable the interface for the mitigation
2510 l1tf= [X86] Control mitigation of the L1TF vulnerability on
2513 The kernel PTE inversion protection is unconditionally
2514 enabled and cannot be disabled.
2517 Provides all available mitigations for the
2518 L1TF vulnerability. Disables SMT and
2519 enables all mitigations in the
2520 hypervisors, i.e. unconditional L1D flush.
2522 SMT control and L1D flush control via the
2523 sysfs interface is still possible after
2524 boot. Hypervisors will issue a warning
2525 when the first VM is started in a
2526 potentially insecure configuration,
2527 i.e. SMT enabled or L1D flush disabled.
2530 Same as 'full', but disables SMT and L1D
2531 flush runtime control. Implies the
2532 'nosmt=force' command line option.
2533 (i.e. sysfs control of SMT is disabled.)
2536 Leaves SMT enabled and enables the default
2537 hypervisor mitigation, i.e. conditional
2540 SMT control and L1D flush control via the
2541 sysfs interface is still possible after
2542 boot. Hypervisors will issue a warning
2543 when the first VM is started in a
2544 potentially insecure configuration,
2545 i.e. SMT enabled or L1D flush disabled.
2549 Disables SMT and enables the default
2550 hypervisor mitigation.
2552 SMT control and L1D flush control via the
2553 sysfs interface is still possible after
2554 boot. Hypervisors will issue a warning
2555 when the first VM is started in a
2556 potentially insecure configuration,
2557 i.e. SMT enabled or L1D flush disabled.
2560 Same as 'flush', but hypervisors will not
2561 warn when a VM is started in a potentially
2562 insecure configuration.
2565 Disables hypervisor mitigations and doesn't
2567 It also drops the swap size and available
2568 RAM limit restriction on both hypervisor and
2573 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2579 lapic [X86-32,APIC] Enable the local APIC even if BIOS
2582 lapic= [X86,APIC] Do not use TSC deadline
2583 value for LAPIC timer one-shot implementation. Default
2584 back to the programmable timer unit in the LAPIC.
2585 Format: notscdeadline
2587 lapic_timer_c2_ok [X86,APIC] trust the local apic timer
2590 libata.dma= [LIBATA] DMA control
2591 libata.dma=0 Disable all PATA and SATA DMA
2592 libata.dma=1 PATA and SATA Disk DMA only
2593 libata.dma=2 ATAPI (CDROM) DMA only
2594 libata.dma=4 Compact Flash DMA only
2595 Combinations also work, so libata.dma=3 enables DMA
2596 for disks and CDROMs, but not CFs.
2598 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2599 libata.ignore_hpa=0 keep BIOS limits (default)
2600 libata.ignore_hpa=1 ignore limits, using full disk
2602 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2606 libata.force= [LIBATA] Force configurations. The format is comma-
2607 separated list of "[ID:]VAL" where ID is
2608 PORT[.DEVICE]. PORT and DEVICE are decimal numbers
2609 matching port, link or device. Basically, it matches
2610 the ATA ID string printed on console by libata. If
2611 the whole ID part is omitted, the last PORT and DEVICE
2612 values are used. If ID hasn't been specified yet, the
2613 configuration applies to all ports, links and devices.
2615 If only DEVICE is omitted, the parameter applies to
2616 the port and all links and devices behind it. DEVICE
2617 number of 0 either selects the first device or the
2618 first fan-out link behind PMP device. It does not
2619 select the host link. DEVICE number of 15 selects the
2620 host link and device attached to it.
2622 The VAL specifies the configuration to force. As long
2623 as there's no ambiguity shortcut notation is allowed.
2624 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2625 The following configurations can be forced.
2627 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2628 Any ID with matching PORT is used.
2630 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2632 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2633 udma[/][16,25,33,44,66,100,133] notation is also
2636 * [no]ncq: Turn on or off NCQ.
2638 * [no]ncqtrim: Turn off queued DSM TRIM.
2640 * nohrst, nosrst, norst: suppress hard, soft
2643 * rstonce: only attempt one reset during
2644 hot-unplug link recovery
2646 * dump_id: dump IDENTIFY data.
2648 * atapi_dmadir: Enable ATAPI DMADIR bridge support
2650 * disable: Disable this device.
2652 If there are multiple matching configurations changing
2653 the same attribute, the last one is used.
2655 memblock=debug [KNL] Enable memblock debug messages.
2657 load_ramdisk= [RAM] [Deprecated]
2659 lockd.nlm_grace_period=P [NFS] Assign grace period.
2662 lockd.nlm_tcpport=N [NFS] Assign TCP port.
2665 lockd.nlm_timeout=T [NFS] Assign timeout value.
2668 lockd.nlm_udpport=M [NFS] Assign UDP port.
2671 lockdown= [SECURITY]
2672 { integrity | confidentiality }
2673 Enable the kernel lockdown feature. If set to
2674 integrity, kernel features that allow userland to
2675 modify the running kernel are disabled. If set to
2676 confidentiality, kernel features that allow userland
2677 to extract confidential information from the kernel
2680 locktorture.nreaders_stress= [KNL]
2681 Set the number of locking read-acquisition kthreads.
2682 Defaults to being automatically set based on the
2683 number of online CPUs.
2685 locktorture.nwriters_stress= [KNL]
2686 Set the number of locking write-acquisition kthreads.
2688 locktorture.onoff_holdoff= [KNL]
2689 Set time (s) after boot for CPU-hotplug testing.
2691 locktorture.onoff_interval= [KNL]
2692 Set time (s) between CPU-hotplug operations, or
2693 zero to disable CPU-hotplug testing.
2695 locktorture.shuffle_interval= [KNL]
2696 Set task-shuffle interval (jiffies). Shuffling
2697 tasks allows some CPUs to go into dyntick-idle
2698 mode during the locktorture test.
2700 locktorture.shutdown_secs= [KNL]
2701 Set time (s) after boot system shutdown. This
2702 is useful for hands-off automated testing.
2704 locktorture.stat_interval= [KNL]
2705 Time (s) between statistics printk()s.
2707 locktorture.stutter= [KNL]
2708 Time (s) to stutter testing, for example,
2709 specifying five seconds causes the test to run for
2710 five seconds, wait for five seconds, and so on.
2711 This tests the locking primitive's ability to
2712 transition abruptly to and from idle.
2714 locktorture.torture_type= [KNL]
2715 Specify the locking implementation to test.
2717 locktorture.verbose= [KNL]
2718 Enable additional printk() statements.
2720 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
2723 loglevel= All Kernel Messages with a loglevel smaller than the
2724 console loglevel will be printed to the console. It can
2725 also be changed with klogd or other programs. The
2726 loglevels are defined as follows:
2728 0 (KERN_EMERG) system is unusable
2729 1 (KERN_ALERT) action must be taken immediately
2730 2 (KERN_CRIT) critical conditions
2731 3 (KERN_ERR) error conditions
2732 4 (KERN_WARNING) warning conditions
2733 5 (KERN_NOTICE) normal but significant condition
2734 6 (KERN_INFO) informational
2735 7 (KERN_DEBUG) debug-level messages
2737 log_buf_len=n[KMG] Sets the size of the printk ring buffer,
2738 in bytes. n must be a power of two and greater
2739 than the minimal size. The minimal size is defined
2740 by LOG_BUF_SHIFT kernel config parameter. There is
2741 also CONFIG_LOG_CPU_MAX_BUF_SHIFT config parameter
2742 that allows to increase the default size depending on
2743 the number of CPUs. See init/Kconfig for more details.
2745 logo.nologo [FB] Disables display of the built-in Linux logo.
2746 This may be used to provide more screen space for
2747 kernel log messages and is useful when debugging
2748 kernel boot problems.
2750 lp=0 [LP] Specify parallel ports to use, e.g,
2751 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
2752 lp=reset first parallel port). 'lp=0' disables the
2753 lp=auto printer driver. 'lp=reset' (which can be
2754 specified in addition to the ports) causes
2755 attached printers to be reset. Using
2756 lp=port1,port2,... specifies the parallel ports
2757 to associate lp devices with, starting with
2758 lp0. A port specification may be 'none' to skip
2759 that lp device, or a parport name such as
2760 'parport0'. Specifying 'lp=auto' instead of a
2761 port specification list means that device IDs
2762 from each port should be examined, to see if
2763 an IEEE 1284-compliant printer is attached; if
2764 so, the driver will manage that printer.
2765 See also header of drivers/char/lp.c.
2768 Sets loops_per_jiffy to given constant, thus avoiding
2769 time-consuming boot-time autodetection (up to 250 ms per
2770 CPU). 0 enables autodetection (default). To determine
2771 the correct value for your kernel, boot with normal
2772 autodetection and see what value is printed. Note that
2773 on SMP systems the preset will be applied to all CPUs,
2774 which is likely to cause problems if your CPUs need
2775 significantly divergent settings. An incorrect value
2776 will cause delays in the kernel to be wrong, leading to
2777 unpredictable I/O errors and other breakage. Although
2778 unlikely, in the extreme case this might damage your
2782 Format: <io>,<irq>,<dma>
2784 lsm.debug [SECURITY] Enable LSM initialization debugging output.
2787 [SECURITY] Choose order of LSM initialization. This
2788 overrides CONFIG_LSM, and the "security=" parameter.
2790 machvec= [IA-64] Force the use of a particular machine-vector
2791 (machvec) in a generic kernel.
2792 Example: machvec=hpzx1
2794 machtype= [Loongson] Share the same kernel image file between
2795 different yeeloong laptops.
2796 Example: machtype=lemote-yeeloong-2f-7inch
2798 max_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory greater
2799 than or equal to this physical address is ignored.
2801 maxcpus= [SMP] Maximum number of processors that an SMP kernel
2802 will bring up during bootup. maxcpus=n : n >= 0 limits
2803 the kernel to bring up 'n' processors. Surely after
2804 bootup you can bring up the other plugged cpu by executing
2805 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
2806 only takes effect during system bootup.
2807 While n=0 is a special case, it is equivalent to "nosmp",
2808 which also disables the IO APIC.
2810 max_loop= [LOOP] The number of loop block devices that get
2811 (loop.max_loop) unconditionally pre-created at init time. The default
2812 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
2813 of statically allocating a predefined number, loop
2814 devices can be requested on-demand with the
2815 /dev/loop-control interface.
2817 mce [X86-32] Machine Check Exception
2819 mce=option [X86-64] See Documentation/x86/x86_64/boot-options.rst
2821 md= [HW] RAID subsystems devices and level
2822 See Documentation/admin-guide/md.rst.
2825 Format: <first>,<last>
2826 Specifies range of consoles to be captured by the MDA.
2829 Control mitigation for the Micro-architectural Data
2830 Sampling (MDS) vulnerability.
2832 Certain CPUs are vulnerable to an exploit against CPU
2833 internal buffers which can forward information to a
2834 disclosure gadget under certain conditions.
2836 In vulnerable processors, the speculatively
2837 forwarded data can be used in a cache side channel
2838 attack, to access data to which the attacker does
2839 not have direct access.
2841 This parameter controls the MDS mitigation. The
2844 full - Enable MDS mitigation on vulnerable CPUs
2845 full,nosmt - Enable MDS mitigation and disable
2846 SMT on vulnerable CPUs
2847 off - Unconditionally disable MDS mitigation
2849 On TAA-affected machines, mds=off can be prevented by
2850 an active TAA mitigation as both vulnerabilities are
2851 mitigated with the same mechanism so in order to disable
2852 this mitigation, you need to specify tsx_async_abort=off
2855 Not specifying this option is equivalent to
2858 For details see: Documentation/admin-guide/hw-vuln/mds.rst
2860 mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
2861 Amount of memory to be used in cases as follows:
2864 2 when the kernel is not able to see the whole system memory;
2865 3 memory that lies after 'mem=' boundary is excluded from
2866 the hypervisor, then assigned to KVM guests.
2868 [X86] Work as limiting max address. Use together
2869 with memmap= to avoid physical address space collisions.
2870 Without memmap= PCI devices could be placed at addresses
2871 belonging to unused RAM.
2873 Note that this only takes effects during boot time since
2874 in above case 3, memory may need be hot added after boot
2875 if system memory of hypervisor is not sufficient.
2877 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
2881 [KNL,SH] Allow user to override the default size for
2882 per-device physically contiguous DMA buffers.
2884 memhp_default_state=online/offline
2885 [KNL] Set the initial state for the memory hotplug
2886 onlining policy. If not specified, the default value is
2887 set according to the
2888 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
2890 See Documentation/admin-guide/mm/memory-hotplug.rst.
2892 memmap=exactmap [KNL,X86] Enable setting of an exact
2893 E820 memory map, as specified by the user.
2894 Such memmap=exactmap lines can be constructed based on
2895 BIOS output or other requirements. See the memmap=nn@ss
2898 memmap=nn[KMG]@ss[KMG]
2899 [KNL, X86, MIPS, XTENSA] Force usage of a specific region of memory.
2900 Region of memory to be used is from ss to ss+nn.
2901 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
2902 which limits max address to nn[KMG].
2903 Multiple different regions can be specified,
2906 memmap=100M@2G,100M#3G,1G!1024G
2908 memmap=nn[KMG]#ss[KMG]
2909 [KNL,ACPI] Mark specific memory as ACPI data.
2910 Region of memory to be marked is from ss to ss+nn.
2912 memmap=nn[KMG]$ss[KMG]
2913 [KNL,ACPI] Mark specific memory as reserved.
2914 Region of memory to be reserved is from ss to ss+nn.
2915 Example: Exclude memory from 0x18690000-0x1869ffff
2916 memmap=64K$0x18690000
2918 memmap=0x10000$0x18690000
2919 Some bootloaders may need an escape character before '$',
2920 like Grub2, otherwise '$' and the following number
2923 memmap=nn[KMG]!ss[KMG]
2924 [KNL,X86] Mark specific memory as protected.
2925 Region of memory to be used, from ss to ss+nn.
2926 The memory region may be marked as e820 type 12 (0xc)
2927 and is NVDIMM or ADR memory.
2929 memmap=<size>%<offset>-<oldtype>+<newtype>
2930 [KNL,ACPI] Convert memory within the specified region
2931 from <oldtype> to <newtype>. If "-<oldtype>" is left
2932 out, the whole region will be marked as <newtype>,
2933 even if previously unavailable. If "+<newtype>" is left
2934 out, matching memory will be removed. Types are
2935 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
2936 3 = ACPI, 12 = PRAM.
2938 memory_corruption_check=0/1 [X86]
2939 Some BIOSes seem to corrupt the first 64k of
2940 memory when doing things like suspend/resume.
2941 Setting this option will scan the memory
2942 looking for corruption. Enabling this will
2943 both detect corruption and prevent the kernel
2944 from using the memory being corrupted.
2945 However, its intended as a diagnostic tool; if
2946 repeatable BIOS-originated corruption always
2947 affects the same memory, you can use memmap=
2948 to prevent the kernel from using that memory.
2950 memory_corruption_check_size=size [X86]
2951 By default it checks for corruption in the low
2952 64k, making this memory unavailable for normal
2953 use. Use this parameter to scan for
2954 corruption in more or less memory.
2956 memory_corruption_check_period=seconds [X86]
2957 By default it checks for corruption every 60
2958 seconds. Use this parameter to check at some
2959 other rate. 0 disables periodic checking.
2961 memory_hotplug.memmap_on_memory
2962 [KNL,X86,ARM] Boolean flag to enable this feature.
2963 Format: {on | off (default)}
2964 When enabled, runtime hotplugged memory will
2965 allocate its internal metadata (struct pages)
2966 from the hotadded memory which will allow to
2967 hotadd a lot of memory without requiring
2968 additional memory to do so.
2969 This feature is disabled by default because it
2970 has some implication on large (e.g. GB)
2971 allocations in some configurations (e.g. small
2973 The state of the flag can be read in
2974 /sys/module/memory_hotplug/parameters/memmap_on_memory.
2975 Note that even when enabled, there are a few cases where
2976 the feature is not effective.
2978 This is not compatible with hugetlb_free_vmemmap. If
2979 both parameters are enabled, hugetlb_free_vmemmap takes
2980 precedence over memory_hotplug.memmap_on_memory.
2982 memtest= [KNL,X86,ARM,PPC,RISCV] Enable memtest
2984 default : 0 <disable>
2985 Specifies the number of memtest passes to be
2986 performed. Each pass selects another test
2987 pattern from a given set of patterns. Memtest
2988 fills the memory with this pattern, validates
2989 memory contents and reserves bad memory
2990 regions that are detected.
2992 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
2993 Valid arguments: on, off
2994 Default (depends on kernel configuration option):
2995 on (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=y)
2996 off (CONFIG_AMD_MEM_ENCRYPT_ACTIVE_BY_DEFAULT=n)
2997 mem_encrypt=on: Activate SME
2998 mem_encrypt=off: Do not activate SME
3000 Refer to Documentation/virt/kvm/amd-memory-encryption.rst
3001 for details on when memory encryption can be activated.
3003 mem_sleep_default= [SUSPEND] Default system suspend mode:
3004 s2idle - Suspend-To-Idle
3005 shallow - Power-On Suspend or equivalent (if supported)
3006 deep - Suspend-To-RAM or equivalent (if supported)
3007 See Documentation/admin-guide/pm/sleep-states.rst.
3009 meye.*= [HW] Set MotionEye Camera parameters
3010 See Documentation/admin-guide/media/meye.rst.
3012 mfgpt_irq= [IA-32] Specify the IRQ to use for the
3013 Multi-Function General Purpose Timers on AMD Geode
3016 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3017 the BIOS has incorrectly applied a workaround. TinyBIOS
3018 version 0.98 is known to be affected, 0.99 fixes the
3019 problem by letting the user disable the workaround.
3023 min_addr=nn[KMG] [KNL,BOOT,ia64] All physical memory below this
3024 physical address is ignored.
3026 mini2440= [ARM,HW,KNL]
3027 Format:[0..2][b][c][t]
3029 MINI2440 configuration specification:
3030 0 - The attached screen is the 3.5" TFT
3031 1 - The attached screen is the 7" TFT
3032 2 - The VGA Shield is attached (1024x768)
3033 Leaving out the screen size parameter will not load
3034 the TFT driver, and the framebuffer will be left
3036 b - Enable backlight. The TFT backlight pin will be
3037 linked to the kernel VESA blanking code and a GPIO
3038 LED. This parameter is not necessary when using the
3040 c - Enable the s3c camera interface.
3041 t - Reserved for enabling touchscreen support. The
3042 touchscreen support is not enabled in the mainstream
3043 kernel as of 2.6.30, a preliminary port can be found
3044 in the "bleeding edge" mini2440 support kernel at
3045 https://repo.or.cz/w/linux-2.6/mini2440.git
3048 [X86,PPC,S390,ARM64] Control optional mitigations for
3049 CPU vulnerabilities. This is a set of curated,
3050 arch-independent options, each of which is an
3051 aggregation of existing arch-specific options.
3054 Disable all optional CPU mitigations. This
3055 improves system performance, but it may also
3056 expose users to several CPU vulnerabilities.
3057 Equivalent to: gather_data_sampling=off [X86]
3059 kvm.nx_huge_pages=off [X86]
3062 mmio_stale_data=off [X86]
3063 no_entry_flush [PPC]
3064 no_uaccess_flush [PPC]
3067 nospectre_v1 [X86,PPC]
3068 nospectre_v2 [X86,PPC,S390,ARM64]
3070 spec_store_bypass_disable=off [X86,PPC]
3071 spectre_v2_user=off [X86]
3072 ssbd=force-off [ARM64]
3073 tsx_async_abort=off [X86]
3076 This does not have any effect on
3077 kvm.nx_huge_pages when
3078 kvm.nx_huge_pages=force.
3081 Mitigate all CPU vulnerabilities, but leave SMT
3082 enabled, even if it's vulnerable. This is for
3083 users who don't want to be surprised by SMT
3084 getting disabled across kernel upgrades, or who
3085 have other ways of avoiding SMT-based attacks.
3086 Equivalent to: (default behavior)
3089 Mitigate all CPU vulnerabilities, disabling SMT
3090 if needed. This is for users who always want to
3091 be fully mitigated, even if it means losing SMT.
3092 Equivalent to: l1tf=flush,nosmt [X86]
3093 mds=full,nosmt [X86]
3094 tsx_async_abort=full,nosmt [X86]
3095 mmio_stale_data=full,nosmt [X86]
3096 retbleed=auto,nosmt [X86]
3099 [KNL] When CONFIG_DEBUG_MEMORY_INIT is set, this
3100 parameter allows control of the logging verbosity for
3101 the additional memory initialisation checks. A value
3102 of 0 disables mminit logging and a level of 4 will
3103 log everything. Information is printed at KERN_DEBUG
3104 so loglevel=8 may also need to be specified.
3107 [X86,INTEL] Control mitigation for the Processor
3108 MMIO Stale Data vulnerabilities.
3110 Processor MMIO Stale Data is a class of
3111 vulnerabilities that may expose data after an MMIO
3112 operation. Exposed data could originate or end in
3113 the same CPU buffers as affected by MDS and TAA.
3114 Therefore, similar to MDS and TAA, the mitigation
3115 is to clear the affected CPU buffers.
3117 This parameter controls the mitigation. The
3120 full - Enable mitigation on vulnerable CPUs
3122 full,nosmt - Enable mitigation and disable SMT on
3125 off - Unconditionally disable mitigation
3127 On MDS or TAA affected machines,
3128 mmio_stale_data=off can be prevented by an active
3129 MDS or TAA mitigation as these vulnerabilities are
3130 mitigated with the same mechanism so in order to
3131 disable this mitigation, you need to specify
3132 mds=off and tsx_async_abort=off too.
3134 Not specifying this option is equivalent to
3135 mmio_stale_data=full.
3138 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3141 [KNL] When CONFIG_MODULE_SIG is set, this means that
3142 modules without (valid) signatures will fail to load.
3143 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3144 is always true, so this option does nothing.
3146 module_blacklist= [KNL] Do not load a comma-separated list of
3147 modules. Useful for debugging problem modules.
3150 [MOUSE] Maximum time between finger touching and
3151 leaving touchpad surface for touch to be considered
3152 a tap and be reported as a left button click (for
3153 touchpads working in absolute mode only).
3155 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3156 reporting absolute coordinates, such as tablets
3157 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3158 reporting absolute coordinates, such as tablets
3160 movablecore= [KNL,X86,IA-64,PPC]
3161 Format: nn[KMGTPE] | nn%
3162 This parameter is the complement to kernelcore=, it
3163 specifies the amount of memory used for migratable
3164 allocations. If both kernelcore and movablecore is
3165 specified, then kernelcore will be at *least* the
3166 specified value but may be more. If movablecore on its
3167 own is specified, the administrator must be careful
3168 that the amount of memory usable for all allocations
3171 movable_node [KNL] Boot-time switch to make hotplugable memory
3172 NUMA nodes to be movable. This means that the memory
3173 of such nodes will be usable only for movable
3174 allocations which rules out almost all kernel
3175 allocations. Use with caution!
3177 MTD_Partition= [MTD]
3178 Format: <name>,<region-number>,<size>,<offset>
3180 MTD_Region= [MTD] Format:
3181 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3184 See drivers/mtd/parsers/cmdlinepart.c
3186 multitce=off [PPC] This parameter disables the use of the pSeries
3187 firmware feature for updating multiple TCE entries
3190 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
3192 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
3194 boundary - index of last SLC block on Flex-OneNAND.
3195 The remaining blocks are configured as MLC blocks.
3196 lock - Configure if Flex-OneNAND boundary should be locked.
3197 Once locked, the boundary cannot be changed.
3198 1 indicates lock status, 0 indicates unlock status.
3201 ARM/S3C2412 JIVE boot control
3203 See arch/arm/mach-s3c/mach-jive.c
3205 mtouchusb.raw_coordinates=
3206 [HW] Make the MicroTouch USB driver use raw coordinates
3207 ('y', default) or cooked coordinates ('n')
3209 mtrr_chunk_size=nn[KMG] [X86]
3210 used for mtrr cleanup. It is largest continuous chunk
3211 that could hold holes aka. UC entries.
3213 mtrr_gran_size=nn[KMG] [X86]
3214 Used for mtrr cleanup. It is granularity of mtrr block.
3216 Large value could prevent small alignment from
3219 mtrr_spare_reg_nr=n [X86]
3221 Range: 0,7 : spare reg number
3223 Used for mtrr cleanup. It is spare mtrr entries number.
3224 Set to 2 or more if your graphical card needs more.
3226 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3228 netdev= [NET] Network devices parameters
3229 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3230 Note that mem_start is often overloaded to mean
3231 something different and driver-specific.
3232 This usage is only documented in each driver source
3236 [NETFILTER] Enable connection tracking flow accounting
3237 0 to disable accounting
3238 1 to enable accounting
3241 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3242 See Documentation/admin-guide/nfs/nfsroot.rst.
3244 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3245 See Documentation/admin-guide/nfs/nfsroot.rst.
3247 nfsrootdebug [NFS] enable nfsroot debugging messages.
3248 See Documentation/admin-guide/nfs/nfsroot.rst.
3250 nfs.callback_nr_threads=
3251 [NFSv4] set the total number of threads that the
3252 NFS client will assign to service NFSv4 callback
3255 nfs.callback_tcpport=
3256 [NFS] set the TCP port on which the NFSv4 callback
3257 channel should listen.
3260 [NFS] sets the pathname to the program which is used
3261 to update the NFS client cache entries.
3263 nfs.cache_getent_timeout=
3264 [NFS] sets the timeout after which an attempt to
3265 update a cache entry is deemed to have failed.
3267 nfs.idmap_cache_timeout=
3268 [NFS] set the maximum lifetime for idmapper cache
3272 [NFS] enable 64-bit inode numbers.
3273 If zero, the NFS client will fake up a 32-bit inode
3274 number for the readdir() and stat() syscalls instead
3275 of returning the full 64-bit number.
3276 The default is to return 64-bit inode numbers.
3278 nfs.max_session_cb_slots=
3279 [NFSv4.1] Sets the maximum number of session
3280 slots the client will assign to the callback
3281 channel. This determines the maximum number of
3282 callbacks the client will process in parallel for
3283 a particular server.
3285 nfs.max_session_slots=
3286 [NFSv4.1] Sets the maximum number of session slots
3287 the client will attempt to negotiate with the server.
3288 This limits the number of simultaneous RPC requests
3289 that the client can send to the NFSv4.1 server.
3290 Note that there is little point in setting this
3291 value higher than the max_tcp_slot_table_limit.
3293 nfs.nfs4_disable_idmapping=
3294 [NFSv4] When set to the default of '1', this option
3295 ensures that both the RPC level authentication
3296 scheme and the NFS level operations agree to use
3297 numeric uids/gids if the mount is using the
3298 'sec=sys' security flavour. In effect it is
3299 disabling idmapping, which can make migration from
3300 legacy NFSv2/v3 systems to NFSv4 easier.
3301 Servers that do not support this mode of operation
3302 will be autodetected by the client, and it will fall
3303 back to using the idmapper.
3304 To turn off this behaviour, set the value to '0'.
3306 [NFS4] Specify an additional fixed unique ident-
3307 ification string that NFSv4 clients can insert into
3308 their nfs_client_id4 string. This is typically a
3309 UUID that is generated at system install time.
3311 nfs.send_implementation_id =
3312 [NFSv4.1] Send client implementation identification
3313 information in exchange_id requests.
3314 If zero, no implementation identification information
3316 The default is to send the implementation identification
3319 nfs.recover_lost_locks =
3320 [NFSv4] Attempt to recover locks that were lost due
3321 to a lease timeout on the server. Please note that
3322 doing this risks data corruption, since there are
3323 no guarantees that the file will remain unchanged
3324 after the locks are lost.
3325 If you want to enable the kernel legacy behaviour of
3326 attempting to recover these locks, then set this
3328 The default parameter value of '0' causes the kernel
3329 not to attempt recovery of lost locks.
3331 nfs4.layoutstats_timer =
3332 [NFSv4.2] Change the rate at which the kernel sends
3333 layoutstats to the pNFS metadata server.
3335 Setting this to value to 0 causes the kernel to use
3336 whatever value is the default set by the layout
3337 driver. A non-zero value sets the minimum interval
3338 in seconds between layoutstats transmissions.
3340 nfsd.nfs4_disable_idmapping=
3341 [NFSv4] When set to the default of '1', the NFSv4
3342 server will return only numeric uids and gids to
3343 clients using auth_sys, and will accept numeric uids
3344 and gids from such clients. This is intended to ease
3345 migration from NFSv2/v3.
3347 nmi_backtrace.backtrace_idle [KNL]
3348 Dump stacks even of idle CPUs in response to an
3349 NMI stack-backtrace request.
3351 nmi_debug= [KNL,SH] Specify one or more actions to take
3352 when a NMI is triggered.
3353 Format: [state][,regs][,debounce][,die]
3355 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3356 Format: [panic,][nopanic,][num]
3358 0 - turn hardlockup detector in nmi_watchdog off
3359 1 - turn hardlockup detector in nmi_watchdog on
3360 When panic is specified, panic when an NMI watchdog
3361 timeout occurs (or 'nopanic' to not panic on an NMI
3362 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3363 To disable both hard and soft lockup detectors,
3364 please see 'nowatchdog'.
3365 This is useful when you use a panic=... timeout and
3366 need the box quickly up again.
3368 These settings can be accessed at runtime via
3369 the nmi_watchdog and hardlockup_panic sysctls.
3371 netpoll.carrier_timeout=
3372 [NET] Specifies amount of time (in seconds) that
3373 netpoll should wait for a carrier. By default netpoll
3376 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3377 emulation library even if a 387 maths coprocessor
3380 no5lvl [X86-64] Disable 5-level paging mode. Forces
3381 kernel to use 4-level paging instead.
3383 nofsgsbase [X86] Disables FSGSBASE instructions.
3386 [HW] Never suspend the console
3387 Disable suspending of consoles during suspend and
3388 hibernate operations. Once disabled, debugging
3389 messages can reach various consoles while the rest
3390 of the system is being put to sleep (ie, while
3391 debugging driver suspend/resume hooks). This may
3392 not work reliably with all consoles, but is known
3393 to work with serial and VGA consoles.
3394 To facilitate more flexible debugging, we also add
3395 console_suspend, a printk module parameter to control
3396 it. Users could use console_suspend (usually
3397 /sys/module/printk/parameters/console_suspend) to
3398 turn on/off it dynamically.
3400 novmcoredd [KNL,KDUMP]
3401 Disable device dump. Device dump allows drivers to
3402 append dump data to vmcore so you can collect driver
3403 specified debug info. Drivers can append the data
3404 without any limit and this data is stored in memory,
3405 so this may cause significant memory stress. Disabling
3406 device dump can help save memory but the driver debug
3407 data will be no longer available. This parameter
3408 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
3411 noaliencache [MM, NUMA, SLAB] Disables the allocation of alien
3412 caches in the slab allocator. Saves per-node memory,
3413 but will impact performance.
3417 noaltinstr [S390] Disables alternative instructions patching
3418 (CPU alternatives feature).
3420 noapic [SMP,APIC] Tells the kernel to not make use of any
3421 IOAPICs that may be present in the system.
3423 noautogroup Disable scheduler automatic task group creation.
3425 nobats [PPC] Do not use BATs for mapping kernel lowmem
3426 on "Classic" PPC cores.
3430 noclflush [BUGS=X86] Don't use the CLFLUSH instruction
3432 delayacct [KNL] Enable per-task delay accounting
3434 nodsp [SH] Disable hardware DSP at boot time.
3436 noefi Disable EFI runtime services support.
3438 no_entry_flush [PPC] Don't flush the L1-D cache when entering the kernel.
3443 On X86-32 available only on PAE configured kernels.
3444 noexec=on: enable non-executable mappings (default)
3445 noexec=off: disable non-executable mappings
3448 Disable SMAP (Supervisor Mode Access Prevention)
3449 even if it is supported by processor.
3452 Disable SMEP (Supervisor Mode Execution Prevention)
3453 even if it is supported by processor.
3456 This affects only 32-bit executables.
3457 noexec32=on: enable non-executable mappings (default)
3458 read doesn't imply executable mappings
3459 noexec32=off: disable non-executable mappings
3460 read implies executable mappings
3462 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3464 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3465 register save and restore. The kernel will only save
3466 legacy floating-point registers on task switch.
3468 nohugeiomap [KNL,X86,PPC,ARM64] Disable kernel huge I/O mappings.
3470 nohugevmalloc [PPC] Disable kernel huge vmalloc mappings.
3472 nosmt [KNL,S390] Disable symmetric multithreading (SMT).
3473 Equivalent to smt=1.
3475 [KNL,X86] Disable symmetric multithreading (SMT).
3476 nosmt=force: Force disable SMT, cannot be undone
3477 via the sysfs control file.
3479 nospectre_v1 [X86,PPC] Disable mitigations for Spectre Variant 1
3480 (bounds check bypass). With this option data leaks are
3481 possible in the system.
3483 nospectre_v2 [X86,PPC_FSL_BOOK3E,ARM64] Disable all mitigations for
3484 the Spectre variant 2 (indirect branch prediction)
3485 vulnerability. System may allow data leaks with this
3488 nospec_store_bypass_disable
3489 [HW] Disable all mitigations for the Speculative Store Bypass vulnerability
3492 [PPC] Don't flush the L1-D cache after accessing user data.
3494 noxsave [BUGS=X86] Disables x86 extended register state save
3495 and restore using xsave. The kernel will fallback to
3496 enabling legacy floating-point and sse state.
3498 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
3499 register states. The kernel will fall back to use
3500 xsave to save the states. By using this parameter,
3501 performance of saving the states is degraded because
3502 xsave doesn't support modified optimization while
3503 xsaveopt supports it on xsaveopt enabled systems.
3505 noxsaves [X86] Disables xsaves and xrstors used in saving and
3506 restoring x86 extended register state in compacted
3507 form of xsave area. The kernel will fall back to use
3508 xsaveopt and xrstor to save and restore the states
3509 in standard form of xsave area. By using this
3510 parameter, xsave area per process might occupy more
3511 memory on xsaves enabled systems.
3513 nohlt [ARM,ARM64,MICROBLAZE,SH] Forces the kernel to busy wait
3514 in do_idle() and not use the arch_cpu_idle()
3515 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3516 to be effective. This is useful on platforms where the
3517 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3518 correctly or when doing power measurements to evalute
3519 the impact of the sleep instructions. This is also
3520 useful when using JTAG debugger.
3522 no_file_caps Tells the kernel not to honor file capabilities. The
3523 only way then for a file to be executed with privilege
3524 is to be setuid root or executed by root.
3526 nohalt [IA-64] Tells the kernel not to use the power saving
3527 function PAL_HALT_LIGHT when idle. This increases
3528 power-consumption. On the positive side, it reduces
3529 interrupt wake-up latency, which may improve performance
3530 in certain environments such as networked servers or
3534 Force pointers printed to the console or buffers to be
3535 unhashed. By default, when a pointer is printed via %p
3536 format string, that pointer is "hashed", i.e. obscured
3537 by hashing the pointer value. This is a security feature
3538 that hides actual kernel addresses from unprivileged
3539 users, but it also makes debugging the kernel more
3540 difficult since unequal pointers can no longer be
3541 compared. However, if this command-line option is
3542 specified, then all normal pointers will have their true
3543 value printed. This option should only be specified when
3544 debugging the kernel. Please do not use on production
3547 nohibernate [HIBERNATION] Disable hibernation and resume.
3549 nohz= [KNL] Boottime enable/disable dynamic ticks
3550 Valid arguments: on, off
3553 nohz_full= [KNL,BOOT,SMP,ISOL]
3554 The argument is a cpu list, as described above.
3555 In kernels built with CONFIG_NO_HZ_FULL=y, set
3556 the specified list of CPUs whose tick will be stopped
3557 whenever possible. The boot CPU will be forced outside
3558 the range to maintain the timekeeping. Any CPUs
3559 in this list will have their RCU callbacks offloaded,
3560 just as if they had also been called out in the
3561 rcu_nocbs= boot parameter.
3563 noiotrap [SH] Disables trapped I/O port accesses.
3565 noirqdebug [X86-32] Disables the code which attempts to detect and
3566 disable unhandled interrupt sources.
3568 no_timer_check [X86,APIC] Disables the code which tests for
3569 broken timer IRQ sources.
3571 noisapnp [ISAPNP] Disables ISA PnP code.
3573 noinitrd [RAM] Tells the kernel not to load any configured
3576 nointremap [X86-64, Intel-IOMMU] Do not enable interrupt
3578 [Deprecated - use intremap=off]
3582 noinvpcid [X86] Disable the INVPCID cpu feature.
3584 nojitter [IA-64] Disables jitter checking for ITC timers.
3586 no-kvmclock [X86,KVM] Disable paravirtualized KVM clock driver
3588 no-kvmapf [X86,KVM] Disable paravirtualized asynchronous page
3592 [X86,PV_OPS] Disable paravirtualized VMware scheduler
3593 clock and use the default one.
3595 no-steal-acc [X86,PV_OPS,ARM64] Disable paravirtualized steal time
3596 accounting. steal time is computed, but won't
3597 influence scheduler behaviour
3599 nolapic [X86-32,APIC] Do not enable or use the local APIC.
3601 nolapic_timer [X86-32,APIC] Do not use the local APIC timer.
3603 noltlbs [PPC] Do not use large page/tlb entries for kernel
3604 lowmem mapping on PPC40x and PPC8xx
3606 nomca [IA-64] Disable machine check abort handling
3608 nomce [X86-32] Disable Machine Check Exception
3610 nomfgpt [X86-32] Disable Multi-Function General Purpose
3611 Timer usage (for AMD Geode machines).
3613 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3614 shutdown the other cpus. Instead use the REBOOT_VECTOR
3617 nomodule Disable module load
3619 nopat [X86] Disable PAT (page attribute table extension of
3620 pagetables) support.
3622 nopcid [X86-64] Disable the PCID cpu feature.
3624 norandmaps Don't use address space randomization. Equivalent to
3625 echo 0 > /proc/sys/kernel/randomize_va_space
3627 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
3628 with UP alternatives
3630 nordrand [X86] Disable kernel use of the RDRAND and
3631 RDSEED instructions even if they are supported
3632 by the processor. RDRAND and RDSEED are still
3633 available to user space applications.
3635 noresume [SWSUSP] Disables resume and restores original swap
3638 no-scroll [VGA] Disables scrollback.
3639 This is required for the Braillex ib80-piezo Braille
3640 reader made by F.H. Papenmeier (Germany).
3644 nosep [BUGS=X86-32] Disables x86 SYSENTER/SYSEXIT support.
3646 nosgx [X86-64,SGX] Disables Intel SGX kernel support.
3648 nosmp [SMP] Tells an SMP kernel to act as a UP kernel,
3649 and disable the IO APIC. legacy for "maxcpus=0".
3651 nosoftlockup [KNL] Disable the soft-lockup detector.
3653 nosync [HW,M68K] Disables sync negotiation for all devices.
3655 nowatchdog [KNL] Disable both lockup detectors, i.e.
3656 soft-lockup and NMI watchdog (hard-lockup).
3660 nox2apic [X86-64,APIC] Do not enable x2APIC mode.
3662 cpu0_hotplug [X86] Turn on CPU0 hotplug feature when
3663 CONFIG_BOOTPARAM_HOTPLUG_CPU0 is off.
3664 Some features depend on CPU0. Known dependencies are:
3665 1. Resume from suspend/hibernate depends on CPU0.
3666 Suspend/hibernate will fail if CPU0 is offline and you
3667 need to online CPU0 before suspend/hibernate.
3668 2. PIC interrupts also depend on CPU0. CPU0 can't be
3669 removed if a PIC interrupt is detected.
3670 It's said poweroff/reboot may depend on CPU0 on some
3671 machines although I haven't seen such issues so far
3672 after CPU0 is offline on a few tested machines.
3673 If the dependencies are under your control, you can
3674 turn on cpu0_hotplug.
3676 nps_mtm_hs_ctr= [KNL,ARC]
3677 This parameter sets the maximum duration, in
3678 cycles, each HW thread of the CTOP can run
3679 without interruptions, before HW switches it.
3680 The actual maximum duration is 16 times this
3682 Format: integer between 1 and 255
3685 nptcg= [IA-64] Override max number of concurrent global TLB
3686 purges which is reported from either PAL_VM_SUMMARY or
3689 nr_cpus= [SMP] Maximum number of processors that an SMP kernel
3690 could support. nr_cpus=n : n >= 1 limits the kernel to
3691 support 'n' processors. It could be larger than the
3692 number of already plugged CPU during bootup, later in
3693 runtime you can physically add extra cpu until it reaches
3694 n. So during boot up some boot time memory for per-cpu
3695 variables need be pre-allocated for later physical cpu
3698 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
3700 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86] Disable NUMA, Only
3701 set up a single NUMA node spanning all memory.
3703 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
3705 Allowed values are enable and disable
3707 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
3708 'node', 'default' can be specified
3709 This can be set from sysctl after boot.
3710 See Documentation/admin-guide/sysctl/vm.rst for details.
3712 ohci1394_dma=early [HW] enable debugging via the ohci1394 driver.
3713 See Documentation/core-api/debugging-via-ohci1394.rst for more
3716 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
3717 Rather than timing out after 20 ms if an EC
3718 command is not properly ACKed, override the length
3719 of the timeout. We have interrupts disabled while
3720 waiting for the ACK, so if this is set too high
3721 interrupts *may* be lost!
3723 omap_mux= [OMAP] Override bootloader pin multiplexing.
3724 Format: <mux_mode0.mode_name=value>...
3725 For example, to override I2C bus2:
3726 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
3728 oops=panic Always panic on oopses. Default is to just kill the
3729 process, but there is a small probability of
3730 deadlocking the machine.
3731 This will also cause panics on machine check exceptions.
3732 Useful together with panic=30 to trigger a reboot.
3735 [KNL] Boolean flag to control whether the page allocator
3736 should randomize its free lists. The randomization may
3737 be automatically enabled if the kernel detects it is
3738 running on a platform with a direct-mapped memory-side
3739 cache, and this parameter can be used to
3740 override/disable that behavior. The state of the flag
3741 can be read from sysfs at:
3742 /sys/module/page_alloc/parameters/shuffle.
3744 page_owner= [KNL] Boot-time page_owner enabling option.
3745 Storage of the information about who allocated
3746 each page is disabled in default. With this switch,
3748 on: enable the feature
3750 page_poison= [KNL] Boot-time parameter changing the state of
3751 poisoning on the buddy allocator, available with
3752 CONFIG_PAGE_POISONING=y.
3753 off: turn off poisoning (default)
3754 on: turn on poisoning
3756 page_reporting.page_reporting_order=
3757 [KNL] Minimal page reporting order
3759 Adjust the minimal page reporting order. The page
3760 reporting is disabled when it exceeds (MAX_ORDER-1).
3762 panic= [KNL] Kernel behaviour on panic: delay <timeout>
3763 timeout > 0: seconds before rebooting
3764 timeout = 0: wait forever
3765 timeout < 0: reboot immediately
3768 panic_print= Bitmask for printing system info when panic happens.
3769 User can chose combination of the following bits:
3770 bit 0: print all tasks info
3771 bit 1: print system memory info
3772 bit 2: print timer info
3773 bit 3: print locks info if CONFIG_LOCKDEP is on
3774 bit 4: print ftrace buffer
3775 bit 5: print all printk messages in buffer
3777 panic_on_taint= Bitmask for conditionally calling panic() in add_taint()
3778 Format: <hex>[,nousertaint]
3779 Hexadecimal bitmask representing the set of TAINT flags
3780 that will cause the kernel to panic when add_taint() is
3781 called with any of the flags in this set.
3782 The optional switch "nousertaint" can be utilized to
3783 prevent userspace forced crashes by writing to sysctl
3784 /proc/sys/kernel/tainted any flagset matching with the
3785 bitmask set on panic_on_taint.
3786 See Documentation/admin-guide/tainted-kernels.rst for
3787 extra details on the taint flags that users can pick
3788 to compose the bitmask to assign to panic_on_taint.
3790 panic_on_warn panic() instead of WARN(). Useful to cause kdump
3793 crash_kexec_post_notifiers
3794 Run kdump after running panic-notifiers and dumping
3795 kmsg. This only for the users who doubt kdump always
3796 succeeds in any situation.
3797 Note that this also increases risks of kdump failure,
3798 because some panic notifiers can make the crashed
3799 kernel more unstable.
3801 parkbd.port= [HW] Parallel port number the keyboard adapter is
3802 connected to, default is 0.
3804 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
3805 0 for XT, 1 for AT (default is AT).
3808 parport= [HW,PPT] Specify parallel ports. 0 disables.
3809 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
3810 Use 'auto' to force the driver to use any
3811 IRQ/DMA settings detected (the default is to
3812 ignore detected IRQ/DMA settings because of
3813 possible conflicts). You can specify the base
3814 address, IRQ, and DMA settings; IRQ and DMA
3815 should be numbers, or 'auto' (for using detected
3816 settings on that particular port), or 'nofifo'
3817 (to avoid using a FIFO even if it is detected).
3818 Parallel ports are assigned in the order they
3819 are specified on the command line, starting
3822 parport_init_mode= [HW,PPT]
3823 Configure VIA parallel port to operate in
3824 a specific mode. This is necessary on Pegasos
3825 computer where firmware has no options for setting
3826 up parallel port mode and sets it to spp.
3827 Currently this function knows 686a and 8231 chips.
3828 Format: [spp|ps2|epp|ecp|ecpepp]
3830 pata_legacy.all= [HW,LIBATA]
3832 Set to non-zero to probe primary and secondary ISA
3833 port ranges on PCI systems where no PCI PATA device
3834 has been found at either range. Disabled by default.
3836 pata_legacy.autospeed= [HW,LIBATA]
3838 Set to non-zero if a chip is present that snoops speed
3839 changes. Disabled by default.
3841 pata_legacy.ht6560a= [HW,LIBATA]
3843 Set to 1, 2, or 3 for HT 6560A on the primary channel,
3844 the secondary channel, or both channels respectively.
3845 Disabled by default.
3847 pata_legacy.ht6560b= [HW,LIBATA]
3849 Set to 1, 2, or 3 for HT 6560B on the primary channel,
3850 the secondary channel, or both channels respectively.
3851 Disabled by default.
3853 pata_legacy.iordy_mask= [HW,LIBATA]
3855 IORDY enable mask. Set individual bits to allow IORDY
3856 for the respective channel. Bit 0 is for the first
3857 legacy channel handled by this driver, bit 1 is for
3858 the second channel, and so on. The sequence will often
3859 correspond to the primary legacy channel, the secondary
3860 legacy channel, and so on, but the handling of a PCI
3861 bus and the use of other driver options may interfere
3862 with the sequence. By default IORDY is allowed across
3865 pata_legacy.opti82c46x= [HW,LIBATA]
3867 Set to 1, 2, or 3 for Opti 82c611A on the primary
3868 channel, the secondary channel, or both channels
3869 respectively. Disabled by default.
3871 pata_legacy.opti82c611a= [HW,LIBATA]
3873 Set to 1, 2, or 3 for Opti 82c465MV on the primary
3874 channel, the secondary channel, or both channels
3875 respectively. Disabled by default.
3877 pata_legacy.pio_mask= [HW,LIBATA]
3879 PIO mode mask for autospeed devices. Set individual
3880 bits to allow the use of the respective PIO modes.
3881 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
3882 All modes allowed by default.
3884 pata_legacy.probe_all= [HW,LIBATA]
3886 Set to non-zero to probe tertiary and further ISA
3887 port ranges on PCI systems. Disabled by default.
3889 pata_legacy.probe_mask= [HW,LIBATA]
3891 Probe mask for legacy ISA PATA ports. Depending on
3892 platform configuration and the use of other driver
3893 options up to 6 legacy ports are supported: 0x1f0,
3894 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
3895 of individual ports can be disabled by setting the
3896 corresponding bits in the mask to 1. Bit 0 is for
3897 the first port in the list above (0x1f0), and so on.
3898 By default all supported ports are probed.
3900 pata_legacy.qdi= [HW,LIBATA]
3902 Set to non-zero to probe QDI controllers. By default
3903 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
3905 pata_legacy.winbond= [HW,LIBATA]
3907 Set to non-zero to probe Winbond controllers. Use
3908 the standard I/O port (0x130) if 1, otherwise the
3909 value given is the I/O port to use (typically 0x1b0).
3910 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
3913 pata_platform.pio_mask= [HW,LIBATA]
3915 Supported PIO mode mask. Set individual bits to allow
3916 the use of the respective PIO modes. Bit 0 is for
3917 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
3921 Halt all CPUs after the first oops has been printed for
3922 the specified number of seconds. This is to be used if
3923 your oopses keep scrolling off the screen.
3928 See header of drivers/block/paride/pcd.c.
3929 See also Documentation/admin-guide/blockdev/paride.rst.
3931 pci=option[,option...] [PCI] various PCI subsystem options.
3933 Some options herein operate on a specific device
3934 or a set of devices (<pci_dev>). These are
3935 specified in one of the following formats:
3937 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
3938 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
3940 Note: the first format specifies a PCI
3941 bus/device/function address which may change
3942 if new hardware is inserted, if motherboard
3943 firmware changes, or due to changes caused
3944 by other kernel parameters. If the
3945 domain is left unspecified, it is
3946 taken to be zero. Optionally, a path
3947 to a device through multiple device/function
3948 addresses can be specified after the base
3949 address (this is more robust against
3950 renumbering issues). The second format
3951 selects devices using IDs from the
3952 configuration space which may match multiple
3953 devices in the system.
3955 earlydump dump PCI config space before the kernel
3957 off [X86] don't probe for the PCI bus
3958 bios [X86-32] force use of PCI BIOS, don't access
3959 the hardware directly. Use this if your machine
3960 has a non-standard PCI host bridge.
3961 nobios [X86-32] disallow use of PCI BIOS, only direct
3962 hardware access methods are allowed. Use this
3963 if you experience crashes upon bootup and you
3964 suspect they are caused by the BIOS.
3965 conf1 [X86] Force use of PCI Configuration Access
3966 Mechanism 1 (config address in IO port 0xCF8,
3967 data in IO port 0xCFC, both 32-bit).
3968 conf2 [X86] Force use of PCI Configuration Access
3969 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
3970 the function, IO port 0xCFA, also 8-bit, sets
3971 bus number. The config space is then accessed
3972 through ports 0xC000-0xCFFF).
3973 See http://wiki.osdev.org/PCI for more info
3974 on the configuration access mechanisms.
3975 noaer [PCIE] If the PCIEAER kernel config parameter is
3976 enabled, this kernel boot option can be used to
3977 disable the use of PCIE advanced error reporting.
3978 nodomains [PCI] Disable support for multiple PCI
3979 root domains (aka PCI segments, in ACPI-speak).
3980 nommconf [X86] Disable use of MMCONFIG for PCI
3982 check_enable_amd_mmconf [X86] check for and enable
3983 properly configured MMIO access to PCI
3984 config space on AMD family 10h CPU
3985 nomsi [MSI] If the PCI_MSI kernel config parameter is
3986 enabled, this kernel boot option can be used to
3987 disable the use of MSI interrupts system-wide.
3988 noioapicquirk [APIC] Disable all boot interrupt quirks.
3989 Safety option to keep boot IRQs enabled. This
3990 should never be necessary.
3991 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
3992 primary IO-APIC for bridges that cannot disable
3993 boot IRQs. This fixes a source of spurious IRQs
3994 when the system masks IRQs.
3995 noioapicreroute [APIC] Disable workaround that uses the
3996 boot IRQ equivalent of an IRQ that connects to
3997 a chipset where boot IRQs cannot be disabled.
3998 The opposite of ioapicreroute.
3999 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4000 routing table. These calls are known to be buggy
4001 on several machines and they hang the machine
4002 when used, but on other computers it's the only
4003 way to get the interrupt routing table. Try
4004 this option if the kernel is unable to allocate
4005 IRQs or discover secondary PCI buses on your
4007 rom [X86] Assign address space to expansion ROMs.
4008 Use with caution as certain devices share
4009 address decoders between ROMs and other
4011 norom [X86] Do not assign address space to
4012 expansion ROMs that do not already have
4013 BIOS assigned address ranges.
4014 nobar [X86] Do not assign address space to the
4015 BARs that weren't assigned by the BIOS.
4016 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4017 assigned automatically to PCI devices. You can
4018 make the kernel exclude IRQs of your ISA cards
4020 pirqaddr=0xAAAAA [X86] Specify the physical address
4021 of the PIRQ table (normally generated
4022 by the BIOS) if it is outside the
4023 F0000h-100000h range.
4024 lastbus=N [X86] Scan all buses thru bus #N. Can be
4025 useful if the kernel is unable to find your
4026 secondary buses and you want to tell it
4027 explicitly which ones they are.
4028 assign-busses [X86] Always assign all PCI bus
4029 numbers ourselves, overriding
4030 whatever the firmware may have done.
4031 usepirqmask [X86] Honor the possible IRQ mask stored
4032 in the BIOS $PIR table. This is needed on
4033 some systems with broken BIOSes, notably
4034 some HP Pavilion N5400 and Omnibook XE3
4035 notebooks. This will have no effect if ACPI
4036 IRQ routing is enabled.
4037 noacpi [X86] Do not use ACPI for IRQ routing
4038 or for PCI scanning.
4039 use_crs [X86] Use PCI host bridge window information
4040 from ACPI. On BIOSes from 2008 or later, this
4041 is enabled by default. If you need to use this,
4042 please report a bug.
4043 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4044 If you need to use this, please report a bug.
4045 routeirq Do IRQ routing for all PCI devices.
4046 This is normally done in pci_enable_device(),
4047 so this option is a temporary workaround
4048 for broken drivers that don't call it.
4049 skip_isa_align [X86] do not align io start addr, so can
4050 handle more pci cards
4051 noearly [X86] Don't do any early type 1 scanning.
4052 This might help on some broken boards which
4053 machine check when some devices' config space
4054 is read. But various workarounds are disabled
4055 and some IOMMU drivers will not work.
4056 bfsort Sort PCI devices into breadth-first order.
4057 This sorting is done to get a device
4058 order compatible with older (<= 2.4) kernels.
4059 nobfsort Don't sort PCI devices into breadth-first order.
4060 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4061 tuning and use the BIOS-configured MPS defaults.
4062 pcie_bus_safe Set every device's MPS to the largest value
4063 supported by all devices below the root complex.
4064 pcie_bus_perf Set device MPS to the largest allowable MPS
4065 based on its parent bus. Also set MRRS (Max
4066 Read Request Size) to the largest supported
4067 value (no larger than the MPS that the device
4068 or bus can support) for best performance.
4069 pcie_bus_peer2peer Set every device's MPS to 128B, which
4070 every device is guaranteed to support. This
4071 configuration allows peer-to-peer DMA between
4072 any pair of devices, possibly at the cost of
4073 reduced performance. This also guarantees
4074 that hot-added devices will work.
4075 cbiosize=nn[KMG] The fixed amount of bus space which is
4076 reserved for the CardBus bridge's IO window.
4077 The default value is 256 bytes.
4078 cbmemsize=nn[KMG] The fixed amount of bus space which is
4079 reserved for the CardBus bridge's memory
4080 window. The default value is 64 megabytes.
4083 [<order of align>@]<pci_dev>[; ...]
4084 Specifies alignment and device to reassign
4085 aligned memory resources. How to
4086 specify the device is described above.
4087 If <order of align> is not specified,
4088 PAGE_SIZE is used as alignment.
4089 A PCI-PCI bridge can be specified if resource
4090 windows need to be expanded.
4091 To specify the alignment for several
4092 instances of a device, the PCI vendor,
4093 device, subvendor, and subdevice may be
4094 specified, e.g., 12@pci:8086:9c22:103c:198f
4095 for 4096-byte alignment.
4096 ecrc= Enable/disable PCIe ECRC (transaction layer
4097 end-to-end CRC checking).
4098 bios: Use BIOS/firmware settings. This is the
4102 hpiosize=nn[KMG] The fixed amount of bus space which is
4103 reserved for hotplug bridge's IO window.
4104 Default size is 256 bytes.
4105 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4106 reserved for hotplug bridge's MMIO window.
4107 Default size is 2 megabytes.
4108 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4109 reserved for hotplug bridge's MMIO_PREF window.
4110 Default size is 2 megabytes.
4111 hpmemsize=nn[KMG] The fixed amount of bus space which is
4112 reserved for hotplug bridge's MMIO and
4114 Default size is 2 megabytes.
4115 hpbussize=nn The minimum amount of additional bus numbers
4116 reserved for buses below a hotplug bridge.
4118 realloc= Enable/disable reallocating PCI bridge resources
4119 if allocations done by BIOS are too small to
4120 accommodate resources required by all child
4122 off: Turn realloc off
4124 realloc same as realloc=on
4125 noari do not use PCIe ARI.
4126 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4127 do not use PCIe ATS (and IOMMU device IOTLB).
4128 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4129 only look for one device below a PCIe downstream
4131 big_root_window Try to add a big 64bit memory window to the PCIe
4132 root complex on AMD CPUs. Some GFX hardware
4133 can resize a BAR to allow access to all VRAM.
4134 Adding the window is slightly risky (it may
4135 conflict with unreported devices), so this
4137 disable_acs_redir=<pci_dev>[; ...]
4138 Specify one or more PCI devices (in the format
4139 specified above) separated by semicolons.
4140 Each device specified will have the PCI ACS
4141 redirect capabilities forced off which will
4142 allow P2P traffic between devices through
4143 bridges without forcing it upstream. Note:
4144 this removes isolation between devices and
4145 may put more devices in an IOMMU group.
4146 force_floating [S390] Force usage of floating interrupts.
4147 nomio [S390] Do not use MIO instructions.
4148 norid [S390] ignore the RID field and force use of
4149 one PCI domain per PCI function
4151 pcie_aspm= [PCIE] Forcibly enable or disable PCIe Active State Power
4154 force Enable ASPM even on devices that claim not to support it.
4155 WARNING: Forcing ASPM on may cause system lockups.
4157 pcie_ports= [PCIE] PCIe port services handling:
4158 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4159 even if the platform doesn't give the OS permission to
4160 use them. This may cause conflicts if the platform
4161 also tries to use these services.
4162 dpc-native Use native PCIe service for DPC only. May
4163 cause conflicts if firmware uses AER or DPC.
4164 compat Disable native PCIe services (PME, AER, DPC, PCIe
4167 pcie_port_pm= [PCIE] PCIe port power management handling:
4168 off Disable power management of all PCIe ports
4169 force Forcibly enable power management of all PCIe ports
4171 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4172 nomsi Do not use MSI for native PCIe PME signaling (this makes
4173 all PCIe root ports use INTx for all services).
4175 pcmv= [HW,PCMCIA] BadgePAD 4
4179 Keep all power-domains already enabled by bootloader on,
4180 even if no driver has claimed them. This is useful
4181 for debug and development, but should not be
4182 needed on a platform with proper driver support.
4185 See Documentation/admin-guide/blockdev/paride.rst.
4187 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4190 See arch/parisc/kernel/pdc_chassis.c
4192 percpu_alloc= Select which percpu first chunk allocator to use.
4193 Currently supported values are "embed" and "page".
4194 Archs may support subset or none of the selections.
4195 See comments in mm/percpu.c for details on each
4196 allocator. This parameter is primarily for debugging
4197 and performance comparison.
4200 See Documentation/admin-guide/blockdev/paride.rst.
4203 See Documentation/admin-guide/blockdev/paride.rst.
4205 pirq= [SMP,APIC] Manual mp-table setup
4206 See Documentation/x86/i386/IO-APIC.rst.
4208 plip= [PPT,NET] Parallel port network link
4209 Format: { parport<nr> | timid | 0 }
4210 See also Documentation/admin-guide/parport.rst.
4212 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4213 Override pmtimer IOPort with a hex value.
4216 pm_debug_messages [SUSPEND,KNL]
4217 Enable suspend/resume debug messages during boot up.
4220 Enable PNP debug messages (depends on the
4221 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4222 via /sys/module/pnp/parameters/debug. We always show
4223 current resource usage; turning this on also shows
4224 possible settings and some assignment information.
4230 { on | off | curr | res | no-curr | no-res }
4233 [ISAPNP] Exclude IRQs for the autoconfiguration
4236 [ISAPNP] Exclude DMAs for the autoconfiguration
4238 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4239 Ranges are in pairs (I/O port base and size).
4242 [ISAPNP] Exclude memory regions for the
4244 Ranges are in pairs (memory base and size).
4246 ports= [IP_VS_FTP] IPVS ftp helper module
4248 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4250 Format: <port>,<port>....
4252 powersave=off [PPC] This option disables power saving features.
4253 It specifically disables cpuidle and sets the
4254 platform machine description specific power_save
4255 function to NULL. On Idle the CPU just reduces
4258 ppc_strict_facility_enable
4259 [PPC] This option catches any kernel floating point,
4260 Altivec, VSX and SPE outside of regions specifically
4261 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4262 There is some performance impact when enabling this.
4266 Disable Hardware Transactional Memory
4269 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4270 none - Limited to cond_resched() calls
4271 voluntary - Limited to cond_resched() and might_sleep() calls
4272 full - Any section that isn't explicitly preempt disabled
4273 can be preempted anytime.
4275 print-fatal-signals=
4276 [KNL] debug: print fatal signals
4278 If enabled, warn about various signal handling
4279 related application anomalies: too many signals,
4280 too many POSIX.1 timers, fatal signals causing a
4283 If you hit the warning due to signal overflow,
4284 you might want to try "ulimit -i unlimited".
4288 printk.always_kmsg_dump=
4289 Trigger kmsg_dump for cases other than kernel oops or
4291 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4294 printk.console_no_auto_verbose=
4295 Disable console loglevel raise on oops, panic
4296 or lockdep-detected issues (only if lock debug is on).
4297 With an exception to setups with low baudrate on
4298 serial console, keeping this 0 is a good choice
4299 in order to provide more debug information.
4301 default: 0 (auto_verbose is enabled)
4303 printk.devkmsg={on,off,ratelimit}
4304 Control writing to /dev/kmsg.
4305 on - unlimited logging to /dev/kmsg from userspace
4306 off - logging to /dev/kmsg disabled
4307 ratelimit - ratelimit the logging
4310 printk.time= Show timing data prefixed to each printk message line
4311 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4313 processor.max_cstate= [HW,ACPI]
4314 Limit processor to maximum C-state
4315 max_cstate=9 overrides any DMI blacklist limit.
4317 processor.nocst [HW,ACPI]
4318 Ignore the _CST method to determine C-states,
4319 instead using the legacy FADT method
4321 profile= [KNL] Enable kernel profiling via /proc/profile
4322 Format: [<profiletype>,]<number>
4323 Param: <profiletype>: "schedule", "sleep", or "kvm"
4324 [defaults to kernel profiling]
4325 Param: "schedule" - profile schedule points.
4326 Param: "sleep" - profile D-state sleeping (millisecs).
4327 Requires CONFIG_SCHEDSTATS
4328 Param: "kvm" - profile VM exits.
4329 Param: <number> - step/bucket size as a power of 2 for
4330 statistical time based profiling.
4332 prompt_ramdisk= [RAM] [Deprecated]
4334 prot_virt= [S390] enable hosting protected virtual machines
4335 isolated from the hypervisor (if hardware supports
4339 psi= [KNL] Enable or disable pressure stall information
4343 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4344 probe for; one of (bare|imps|exps|lifebook|any).
4345 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4347 psmouse.resetafter= [HW,MOUSE]
4348 Try to reset the device after so many bad packets
4351 [HW,MOUSE] Set desired mouse resolution, in dpi.
4352 psmouse.smartscroll=
4353 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4354 0 = disabled, 1 = enabled (default).
4356 pstore.backend= Specify the name of the pstore backend to use
4359 See Documentation/admin-guide/blockdev/paride.rst.
4361 pti= [X86-64] Control Page Table Isolation of user and
4362 kernel address spaces. Disabling this feature
4363 removes hardening, but improves performance of
4364 system calls and interrupts.
4366 on - unconditionally enable
4367 off - unconditionally disable
4368 auto - kernel detects whether your CPU model is
4369 vulnerable to issues that PTI mitigates
4371 Not specifying this option is equivalent to pti=auto.
4374 Equivalent to pti=off
4377 [KNL] Number of legacy pty's. Overwrites compiled-in
4380 quiet [KNL] Disable most log messages
4385 See Documentation/admin-guide/md.rst.
4387 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4388 See Documentation/admin-guide/blockdev/ramdisk.rst.
4390 ramdisk_start= [RAM] RAM disk image start address
4392 random.trust_cpu={on,off}
4393 [KNL] Enable or disable trusting the use of the
4394 CPU's random number generator (if available) to
4395 fully seed the kernel's CRNG. Default is controlled
4396 by CONFIG_RANDOM_TRUST_CPU.
4398 random.trust_bootloader={on,off}
4399 [KNL] Enable or disable trusting the use of a
4400 seed passed by the bootloader (if available) to
4401 fully seed the kernel's CRNG. Default is controlled
4402 by CONFIG_RANDOM_TRUST_BOOTLOADER.
4404 randomize_kstack_offset=
4405 [KNL] Enable or disable kernel stack offset
4406 randomization, which provides roughly 5 bits of
4407 entropy, frustrating memory corruption attacks
4408 that depend on stack address determinism or
4409 cross-syscall address exposures. This is only
4410 available on architectures that have defined
4411 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4412 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4413 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4415 ras=option[,option,...] [KNL] RAS-specific options
4418 Disable the Correctable Errors Collector,
4419 see CONFIG_RAS_CEC help text.
4422 The argument is a cpu list, as described above.
4424 In kernels built with CONFIG_RCU_NOCB_CPU=y, set
4425 the specified list of CPUs to be no-callback CPUs.
4426 Invocation of these CPUs' RCU callbacks will be
4427 offloaded to "rcuox/N" kthreads created for that
4428 purpose, where "x" is "p" for RCU-preempt, and
4429 "s" for RCU-sched, and "N" is the CPU number.
4430 This reduces OS jitter on the offloaded CPUs,
4431 which can be useful for HPC and real-time
4432 workloads. It can also improve energy efficiency
4433 for asymmetric multiprocessors.
4436 Rather than requiring that offloaded CPUs
4437 (specified by rcu_nocbs= above) explicitly
4438 awaken the corresponding "rcuoN" kthreads,
4439 make these kthreads poll for callbacks.
4440 This improves the real-time response for the
4441 offloaded CPUs by relieving them of the need to
4442 wake up the corresponding kthread, but degrades
4443 energy efficiency by requiring that the kthreads
4444 periodically wake up to do the polling.
4446 rcutree.blimit= [KNL]
4447 Set maximum number of finished RCU callbacks to
4448 process in one batch.
4450 rcutree.dump_tree= [KNL]
4451 Dump the structure of the rcu_node combining tree
4452 out at early boot. This is used for diagnostic
4453 purposes, to verify correct tree setup.
4455 rcutree.gp_cleanup_delay= [KNL]
4456 Set the number of jiffies to delay each step of
4457 RCU grace-period cleanup.
4459 rcutree.gp_init_delay= [KNL]
4460 Set the number of jiffies to delay each step of
4461 RCU grace-period initialization.
4463 rcutree.gp_preinit_delay= [KNL]
4464 Set the number of jiffies to delay each step of
4465 RCU grace-period pre-initialization, that is,
4466 the propagation of recent CPU-hotplug changes up
4467 the rcu_node combining tree.
4469 rcutree.use_softirq= [KNL]
4470 If set to zero, move all RCU_SOFTIRQ processing to
4471 per-CPU rcuc kthreads. Defaults to a non-zero
4472 value, meaning that RCU_SOFTIRQ is used by default.
4473 Specify rcutree.use_softirq=0 to use rcuc kthreads.
4475 But note that CONFIG_PREEMPT_RT=y kernels disable
4476 this kernel boot parameter, forcibly setting it
4479 rcutree.rcu_fanout_exact= [KNL]
4480 Disable autobalancing of the rcu_node combining
4481 tree. This is used by rcutorture, and might
4482 possibly be useful for architectures having high
4483 cache-to-cache transfer latencies.
4485 rcutree.rcu_fanout_leaf= [KNL]
4486 Change the number of CPUs assigned to each
4487 leaf rcu_node structure. Useful for very
4488 large systems, which will choose the value 64,
4489 and for NUMA systems with large remote-access
4490 latencies, which will choose a value aligned
4491 with the appropriate hardware boundaries.
4493 rcutree.rcu_min_cached_objs= [KNL]
4494 Minimum number of objects which are cached and
4495 maintained per one CPU. Object size is equal
4496 to PAGE_SIZE. The cache allows to reduce the
4497 pressure to page allocator, also it makes the
4498 whole algorithm to behave better in low memory
4501 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
4502 Set the page-cache refill delay (in milliseconds)
4503 in response to low-memory conditions. The range
4504 of permitted values is in the range 0:100000.
4506 rcutree.jiffies_till_first_fqs= [KNL]
4507 Set delay from grace-period initialization to
4508 first attempt to force quiescent states.
4509 Units are jiffies, minimum value is zero,
4510 and maximum value is HZ.
4512 rcutree.jiffies_till_next_fqs= [KNL]
4513 Set delay between subsequent attempts to force
4514 quiescent states. Units are jiffies, minimum
4515 value is one, and maximum value is HZ.
4517 rcutree.jiffies_till_sched_qs= [KNL]
4518 Set required age in jiffies for a
4519 given grace period before RCU starts
4520 soliciting quiescent-state help from
4521 rcu_note_context_switch() and cond_resched().
4522 If not specified, the kernel will calculate
4523 a value based on the most recent settings
4524 of rcutree.jiffies_till_first_fqs
4525 and rcutree.jiffies_till_next_fqs.
4526 This calculated value may be viewed in
4527 rcutree.jiffies_to_sched_qs. Any attempt to set
4528 rcutree.jiffies_to_sched_qs will be cheerfully
4531 rcutree.kthread_prio= [KNL,BOOT]
4532 Set the SCHED_FIFO priority of the RCU per-CPU
4533 kthreads (rcuc/N). This value is also used for
4534 the priority of the RCU boost threads (rcub/N)
4535 and for the RCU grace-period kthreads (rcu_bh,
4536 rcu_preempt, and rcu_sched). If RCU_BOOST is
4537 set, valid values are 1-99 and the default is 1
4538 (the least-favored priority). Otherwise, when
4539 RCU_BOOST is not set, valid values are 0-99 and
4540 the default is zero (non-realtime operation).
4542 rcutree.rcu_nocb_gp_stride= [KNL]
4543 Set the number of NOCB callback kthreads in
4544 each group, which defaults to the square root
4545 of the number of CPUs. Larger numbers reduce
4546 the wakeup overhead on the global grace-period
4547 kthread, but increases that same overhead on
4548 each group's NOCB grace-period kthread.
4550 rcutree.qhimark= [KNL]
4551 Set threshold of queued RCU callbacks beyond which
4552 batch limiting is disabled.
4554 rcutree.qlowmark= [KNL]
4555 Set threshold of queued RCU callbacks below which
4556 batch limiting is re-enabled.
4558 rcutree.qovld= [KNL]
4559 Set threshold of queued RCU callbacks beyond which
4560 RCU's force-quiescent-state scan will aggressively
4561 enlist help from cond_resched() and sched IPIs to
4562 help CPUs more quickly reach quiescent states.
4563 Set to less than zero to make this be set based
4564 on rcutree.qhimark at boot time and to zero to
4565 disable more aggressive help enlistment.
4567 rcutree.rcu_idle_gp_delay= [KNL]
4568 Set wakeup interval for idle CPUs that have
4569 RCU callbacks (RCU_FAST_NO_HZ=y).
4571 rcutree.rcu_kick_kthreads= [KNL]
4572 Cause the grace-period kthread to get an extra
4573 wake_up() if it sleeps three times longer than
4574 it should at force-quiescent-state time.
4575 This wake_up() will be accompanied by a
4576 WARN_ONCE() splat and an ftrace_dump().
4578 rcutree.rcu_unlock_delay= [KNL]
4579 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
4580 this specifies an rcu_read_unlock()-time delay
4581 in microseconds. This defaults to zero.
4582 Larger delays increase the probability of
4583 catching RCU pointer leaks, that is, buggy use
4584 of RCU-protected pointers after the relevant
4585 rcu_read_unlock() has completed.
4587 rcutree.sysrq_rcu= [KNL]
4588 Commandeer a sysrq key to dump out Tree RCU's
4589 rcu_node tree with an eye towards determining
4590 why a new grace period has not yet started.
4592 rcuscale.gp_async= [KNL]
4593 Measure performance of asynchronous
4594 grace-period primitives such as call_rcu().
4596 rcuscale.gp_async_max= [KNL]
4597 Specify the maximum number of outstanding
4598 callbacks per writer thread. When a writer
4599 thread exceeds this limit, it invokes the
4600 corresponding flavor of rcu_barrier() to allow
4601 previously posted callbacks to drain.
4603 rcuscale.gp_exp= [KNL]
4604 Measure performance of expedited synchronous
4605 grace-period primitives.
4607 rcuscale.holdoff= [KNL]
4608 Set test-start holdoff period. The purpose of
4609 this parameter is to delay the start of the
4610 test until boot completes in order to avoid
4613 rcuscale.kfree_rcu_test= [KNL]
4614 Set to measure performance of kfree_rcu() flooding.
4616 rcuscale.kfree_rcu_test_double= [KNL]
4617 Test the double-argument variant of kfree_rcu().
4618 If this parameter has the same value as
4619 rcuscale.kfree_rcu_test_single, both the single-
4620 and double-argument variants are tested.
4622 rcuscale.kfree_rcu_test_single= [KNL]
4623 Test the single-argument variant of kfree_rcu().
4624 If this parameter has the same value as
4625 rcuscale.kfree_rcu_test_double, both the single-
4626 and double-argument variants are tested.
4628 rcuscale.kfree_nthreads= [KNL]
4629 The number of threads running loops of kfree_rcu().
4631 rcuscale.kfree_alloc_num= [KNL]
4632 Number of allocations and frees done in an iteration.
4634 rcuscale.kfree_loops= [KNL]
4635 Number of loops doing rcuscale.kfree_alloc_num number
4636 of allocations and frees.
4638 rcuscale.nreaders= [KNL]
4639 Set number of RCU readers. The value -1 selects
4640 N, where N is the number of CPUs. A value
4641 "n" less than -1 selects N-n+1, where N is again
4642 the number of CPUs. For example, -2 selects N
4643 (the number of CPUs), -3 selects N+1, and so on.
4644 A value of "n" less than or equal to -N selects
4647 rcuscale.nwriters= [KNL]
4648 Set number of RCU writers. The values operate
4649 the same as for rcuscale.nreaders.
4650 N, where N is the number of CPUs
4652 rcuscale.perf_type= [KNL]
4653 Specify the RCU implementation to test.
4655 rcuscale.shutdown= [KNL]
4656 Shut the system down after performance tests
4657 complete. This is useful for hands-off automated
4660 rcuscale.verbose= [KNL]
4661 Enable additional printk() statements.
4663 rcuscale.writer_holdoff= [KNL]
4664 Write-side holdoff between grace periods,
4665 in microseconds. The default of zero says
4668 rcutorture.fqs_duration= [KNL]
4669 Set duration of force_quiescent_state bursts
4672 rcutorture.fqs_holdoff= [KNL]
4673 Set holdoff time within force_quiescent_state bursts
4676 rcutorture.fqs_stutter= [KNL]
4677 Set wait time between force_quiescent_state bursts
4680 rcutorture.fwd_progress= [KNL]
4681 Enable RCU grace-period forward-progress testing
4682 for the types of RCU supporting this notion.
4684 rcutorture.fwd_progress_div= [KNL]
4685 Specify the fraction of a CPU-stall-warning
4686 period to do tight-loop forward-progress testing.
4688 rcutorture.fwd_progress_holdoff= [KNL]
4689 Number of seconds to wait between successive
4690 forward-progress tests.
4692 rcutorture.fwd_progress_need_resched= [KNL]
4693 Enclose cond_resched() calls within checks for
4694 need_resched() during tight-loop forward-progress
4697 rcutorture.gp_cond= [KNL]
4698 Use conditional/asynchronous update-side
4699 primitives, if available.
4701 rcutorture.gp_exp= [KNL]
4702 Use expedited update-side primitives, if available.
4704 rcutorture.gp_normal= [KNL]
4705 Use normal (non-expedited) asynchronous
4706 update-side primitives, if available.
4708 rcutorture.gp_sync= [KNL]
4709 Use normal (non-expedited) synchronous
4710 update-side primitives, if available. If all
4711 of rcutorture.gp_cond=, rcutorture.gp_exp=,
4712 rcutorture.gp_normal=, and rcutorture.gp_sync=
4713 are zero, rcutorture acts as if is interpreted
4714 they are all non-zero.
4716 rcutorture.irqreader= [KNL]
4717 Run RCU readers from irq handlers, or, more
4718 accurately, from a timer handler. Not all RCU
4719 flavors take kindly to this sort of thing.
4721 rcutorture.leakpointer= [KNL]
4722 Leak an RCU-protected pointer out of the reader.
4723 This can of course result in splats, and is
4724 intended to test the ability of things like
4725 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
4728 rcutorture.n_barrier_cbs= [KNL]
4729 Set callbacks/threads for rcu_barrier() testing.
4731 rcutorture.nfakewriters= [KNL]
4732 Set number of concurrent RCU writers. These just
4733 stress RCU, they don't participate in the actual
4734 test, hence the "fake".
4736 rcutorture.nocbs_nthreads= [KNL]
4737 Set number of RCU callback-offload togglers.
4738 Zero (the default) disables toggling.
4740 rcutorture.nocbs_toggle= [KNL]
4741 Set the delay in milliseconds between successive
4742 callback-offload toggling attempts.
4744 rcutorture.nreaders= [KNL]
4745 Set number of RCU readers. The value -1 selects
4746 N-1, where N is the number of CPUs. A value
4747 "n" less than -1 selects N-n-2, where N is again
4748 the number of CPUs. For example, -2 selects N
4749 (the number of CPUs), -3 selects N+1, and so on.
4751 rcutorture.object_debug= [KNL]
4752 Enable debug-object double-call_rcu() testing.
4754 rcutorture.onoff_holdoff= [KNL]
4755 Set time (s) after boot for CPU-hotplug testing.
4757 rcutorture.onoff_interval= [KNL]
4758 Set time (jiffies) between CPU-hotplug operations,
4759 or zero to disable CPU-hotplug testing.
4761 rcutorture.read_exit= [KNL]
4762 Set the number of read-then-exit kthreads used
4763 to test the interaction of RCU updaters and
4764 task-exit processing.
4766 rcutorture.read_exit_burst= [KNL]
4767 The number of times in a given read-then-exit
4768 episode that a set of read-then-exit kthreads
4771 rcutorture.read_exit_delay= [KNL]
4772 The delay, in seconds, between successive
4773 read-then-exit testing episodes.
4775 rcutorture.shuffle_interval= [KNL]
4776 Set task-shuffle interval (s). Shuffling tasks
4777 allows some CPUs to go into dyntick-idle mode
4778 during the rcutorture test.
4780 rcutorture.shutdown_secs= [KNL]
4781 Set time (s) after boot system shutdown. This
4782 is useful for hands-off automated testing.
4784 rcutorture.stall_cpu= [KNL]
4785 Duration of CPU stall (s) to test RCU CPU stall
4786 warnings, zero to disable.
4788 rcutorture.stall_cpu_block= [KNL]
4789 Sleep while stalling if set. This will result
4790 in warnings from preemptible RCU in addition
4791 to any other stall-related activity.
4793 rcutorture.stall_cpu_holdoff= [KNL]
4794 Time to wait (s) after boot before inducing stall.
4796 rcutorture.stall_cpu_irqsoff= [KNL]
4797 Disable interrupts while stalling if set.
4799 rcutorture.stall_gp_kthread= [KNL]
4800 Duration (s) of forced sleep within RCU
4801 grace-period kthread to test RCU CPU stall
4802 warnings, zero to disable. If both stall_cpu
4803 and stall_gp_kthread are specified, the
4804 kthread is starved first, then the CPU.
4806 rcutorture.stat_interval= [KNL]
4807 Time (s) between statistics printk()s.
4809 rcutorture.stutter= [KNL]
4810 Time (s) to stutter testing, for example, specifying
4811 five seconds causes the test to run for five seconds,
4812 wait for five seconds, and so on. This tests RCU's
4813 ability to transition abruptly to and from idle.
4815 rcutorture.test_boost= [KNL]
4816 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
4817 "Maybe" means test if the RCU implementation
4818 under test support RCU priority boosting.
4820 rcutorture.test_boost_duration= [KNL]
4821 Duration (s) of each individual boost test.
4823 rcutorture.test_boost_interval= [KNL]
4824 Interval (s) between each boost test.
4826 rcutorture.test_no_idle_hz= [KNL]
4827 Test RCU's dyntick-idle handling. See also the
4828 rcutorture.shuffle_interval parameter.
4830 rcutorture.torture_type= [KNL]
4831 Specify the RCU implementation to test.
4833 rcutorture.verbose= [KNL]
4834 Enable additional printk() statements.
4836 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
4837 Dump ftrace buffer after reporting RCU CPU
4840 rcupdate.rcu_cpu_stall_suppress= [KNL]
4841 Suppress RCU CPU stall warning messages.
4843 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
4844 Suppress RCU CPU stall warning messages and
4845 rcutorture writer stall warnings that occur
4846 during early boot, that is, during the time
4847 before the init task is spawned.
4849 rcupdate.rcu_cpu_stall_timeout= [KNL]
4850 Set timeout for RCU CPU stall warning messages.
4852 rcupdate.rcu_expedited= [KNL]
4853 Use expedited grace-period primitives, for
4854 example, synchronize_rcu_expedited() instead
4855 of synchronize_rcu(). This reduces latency,
4856 but can increase CPU utilization, degrade
4857 real-time latency, and degrade energy efficiency.
4858 No effect on CONFIG_TINY_RCU kernels.
4860 rcupdate.rcu_normal= [KNL]
4861 Use only normal grace-period primitives,
4862 for example, synchronize_rcu() instead of
4863 synchronize_rcu_expedited(). This improves
4864 real-time latency, CPU utilization, and
4865 energy efficiency, but can expose users to
4866 increased grace-period latency. This parameter
4867 overrides rcupdate.rcu_expedited. No effect on
4868 CONFIG_TINY_RCU kernels.
4870 rcupdate.rcu_normal_after_boot= [KNL]
4871 Once boot has completed (that is, after
4872 rcu_end_inkernel_boot() has been invoked), use
4873 only normal grace-period primitives. No effect
4874 on CONFIG_TINY_RCU kernels.
4876 But note that CONFIG_PREEMPT_RT=y kernels enables
4877 this kernel boot parameter, forcibly setting
4878 it to the value one, that is, converting any
4879 post-boot attempt at an expedited RCU grace
4880 period to instead use normal non-expedited
4881 grace-period processing.
4883 rcupdate.rcu_task_ipi_delay= [KNL]
4884 Set time in jiffies during which RCU tasks will
4885 avoid sending IPIs, starting with the beginning
4886 of a given grace period. Setting a large
4887 number avoids disturbing real-time workloads,
4888 but lengthens grace periods.
4890 rcupdate.rcu_task_stall_timeout= [KNL]
4891 Set timeout in jiffies for RCU task stall warning
4892 messages. Disable with a value less than or equal
4895 rcupdate.rcu_self_test= [KNL]
4896 Run the RCU early boot self tests
4900 Run specified binary instead of /init from the ramdisk,
4901 used for early userspace startup. See initrd.
4904 force - Override the decision by the kernel to hide the
4905 advertisement of RDRAND support (this affects
4906 certain AMD processors because of buggy BIOS
4907 support, specifically around the suspend/resume
4911 Turn on/off individual RDT features. List is:
4912 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
4914 E.g. to turn on cmt and turn off mba use:
4918 Format (x86 or x86_64):
4919 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
4921 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
4923 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
4924 (prefix with 'panic_' to set mode for panic
4926 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
4927 reboot_force is either force or not specified,
4928 reboot_cpu is s[mp]#### with #### being the processor
4929 to be used for rebooting.
4931 refscale.holdoff= [KNL]
4932 Set test-start holdoff period. The purpose of
4933 this parameter is to delay the start of the
4934 test until boot completes in order to avoid
4937 refscale.loops= [KNL]
4938 Set the number of loops over the synchronization
4939 primitive under test. Increasing this number
4940 reduces noise due to loop start/end overhead,
4941 but the default has already reduced the per-pass
4942 noise to a handful of picoseconds on ca. 2020
4945 refscale.nreaders= [KNL]
4946 Set number of readers. The default value of -1
4947 selects N, where N is roughly 75% of the number
4948 of CPUs. A value of zero is an interesting choice.
4950 refscale.nruns= [KNL]
4951 Set number of runs, each of which is dumped onto
4954 refscale.readdelay= [KNL]
4955 Set the read-side critical-section duration,
4956 measured in microseconds.
4958 refscale.scale_type= [KNL]
4959 Specify the read-protection implementation to test.
4961 refscale.shutdown= [KNL]
4962 Shut down the system at the end of the performance
4963 test. This defaults to 1 (shut it down) when
4964 refscale is built into the kernel and to 0 (leave
4965 it running) when refscale is built as a module.
4967 refscale.verbose= [KNL]
4968 Enable additional printk() statements.
4970 refscale.verbose_batched= [KNL]
4971 Batch the additional printk() statements. If zero
4972 (the default) or negative, print everything. Otherwise,
4973 print every Nth verbose statement, where N is the value
4977 [KNL, SMP] Set scheduler's default relax_domain_level.
4978 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
4980 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
4981 Format: <base1>,<size1>[,<base2>,<size2>,...]
4982 Reserve I/O ports or memory so the kernel won't use
4983 them. If <base> is less than 0x10000, the region
4984 is assumed to be I/O ports; otherwise it is memory.
4986 reservetop= [X86-32]
4988 Reserves a hole at the top of the kernel virtual
4991 reset_devices [KNL] Force drivers to reset the underlying device
4992 during initialization.
4995 Specify the partition device for software suspend
4997 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
4999 resume_offset= [SWSUSP]
5000 Specify the offset from the beginning of the partition
5001 given by "resume=" at which the swap header is located,
5002 in <PAGE_SIZE> units (needed only for swap files).
5003 See Documentation/power/swsusp-and-swap-files.rst
5005 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5006 read the resume files
5008 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5009 Useful for devices that are detected asynchronously
5010 (e.g. USB and MMC devices).
5012 hibernate= [HIBERNATION]
5013 noresume Don't check if there's a hibernation image
5014 present during boot.
5015 nocompress Don't compress/decompress hibernation images.
5016 no Disable hibernation and resume.
5017 protect_image Turn on image protection during restoration
5018 (that will set all pages holding image data
5019 during restoration read-only).
5021 retain_initrd [RAM] Keep initrd memory after extraction
5023 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5024 Speculative Code Execution with Return Instructions)
5027 AMD-based UNRET and IBPB mitigations alone do not stop
5028 sibling threads from influencing the predictions of other
5029 sibling threads. For that reason, STIBP is used on pro-
5030 cessors that support it, and mitigate SMT on processors
5034 auto - automatically select a migitation
5035 auto,nosmt - automatically select a mitigation,
5036 disabling SMT if necessary for
5037 the full mitigation (only on Zen1
5038 and older without STIBP).
5039 ibpb - On AMD, mitigate short speculation
5040 windows on basic block boundaries too.
5041 Safe, highest perf impact. It also
5042 enables STIBP if present. Not suitable
5044 ibpb,nosmt - Like "ibpb" above but will disable SMT
5045 when STIBP is not available. This is
5046 the alternative for systems which do not
5048 unret - Force enable untrained return thunks,
5049 only effective on AMD f15h-f17h based
5051 unret,nosmt - Like unret, but will disable SMT when STIBP
5052 is not available. This is the alternative for
5053 systems which do not have STIBP.
5055 Selecting 'auto' will choose a mitigation method at run
5056 time according to the CPU.
5058 Not specifying this option is equivalent to retbleed=auto.
5060 rfkill.default_state=
5061 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5062 etc. communication is blocked by default.
5065 rfkill.master_switch_mode=
5066 0 The "airplane mode" button does nothing.
5067 1 The "airplane mode" button toggles between everything
5068 blocked and the previous configuration.
5069 2 The "airplane mode" button toggles between everything
5070 blocked and everything unblocked.
5072 rhash_entries= [KNL,NET]
5073 Set number of hash buckets for route cache
5076 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5079 ro [KNL] Mount root device read-only on boot
5082 on Mark read-only kernel memory as read-only (default).
5083 off Leave read-only kernel memory writable for debugging.
5086 Enable the uart passthrough on the designated usb port
5087 on Rockchip SoCs. When active, the signals of the
5088 debug-uart get routed to the D+ and D- pins of the usb
5089 port and the regular usb controller gets disabled.
5091 root= [KNL] Root filesystem
5092 See name_to_dev_t comment in init/do_mounts.c.
5094 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5095 mount the root filesystem
5097 rootflags= [KNL] Set root filesystem mount option string
5099 rootfstype= [KNL] Set root filesystem type
5101 rootwait [KNL] Wait (indefinitely) for root device to show up.
5102 Useful for devices that are detected asynchronously
5103 (e.g. USB and MMC devices).
5105 rproc_mem=nn[KMG][@address]
5106 [KNL,ARM,CMA] Remoteproc physical memory block.
5107 Memory area to be used by remote processor image,
5110 rw [KNL] Mount root device read-write on boot
5112 S [KNL] Run init in single mode
5114 s390_iommu= [HW,S390]
5115 Set s390 IOTLB flushing mode
5117 With strict flushing every unmap operation will result in
5118 an IOTLB flush. Default is lazy flushing before reuse,
5122 See drivers/net/irda/sa1100_ir.c.
5124 sched_verbose [KNL] Enables verbose scheduler debug messages.
5126 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5127 Allowed values are enable and disable. This feature
5128 incurs a small amount of overhead in the scheduler
5129 but is useful for debugging and performance tuning.
5131 sched_thermal_decay_shift=
5132 [KNL, SMP] Set a decay shift for scheduler thermal
5133 pressure signal. Thermal pressure signal follows the
5134 default decay period of other scheduler pelt
5135 signals(usually 32 ms but configurable). Setting
5136 sched_thermal_decay_shift will left shift the decay
5137 period for the thermal pressure signal by the shift
5139 i.e. with the default pelt decay period of 32 ms
5140 sched_thermal_decay_shift thermal pressure decay pr
5144 Format: integer between 0 and 10
5147 scftorture.holdoff= [KNL]
5148 Number of seconds to hold off before starting
5149 test. Defaults to zero for module insertion and
5150 to 10 seconds for built-in smp_call_function()
5153 scftorture.longwait= [KNL]
5154 Request ridiculously long waits randomly selected
5155 up to the chosen limit in seconds. Zero (the
5156 default) disables this feature. Please note
5157 that requesting even small non-zero numbers of
5158 seconds can result in RCU CPU stall warnings,
5159 softlockup complaints, and so on.
5161 scftorture.nthreads= [KNL]
5162 Number of kthreads to spawn to invoke the
5163 smp_call_function() family of functions.
5164 The default of -1 specifies a number of kthreads
5165 equal to the number of CPUs.
5167 scftorture.onoff_holdoff= [KNL]
5168 Number seconds to wait after the start of the
5169 test before initiating CPU-hotplug operations.
5171 scftorture.onoff_interval= [KNL]
5172 Number seconds to wait between successive
5173 CPU-hotplug operations. Specifying zero (which
5174 is the default) disables CPU-hotplug operations.
5176 scftorture.shutdown_secs= [KNL]
5177 The number of seconds following the start of the
5178 test after which to shut down the system. The
5179 default of zero avoids shutting down the system.
5180 Non-zero values are useful for automated tests.
5182 scftorture.stat_interval= [KNL]
5183 The number of seconds between outputting the
5184 current test statistics to the console. A value
5185 of zero disables statistics output.
5187 scftorture.stutter_cpus= [KNL]
5188 The number of jiffies to wait between each change
5189 to the set of CPUs under test.
5191 scftorture.use_cpus_read_lock= [KNL]
5192 Use use_cpus_read_lock() instead of the default
5193 preempt_disable() to disable CPU hotplug
5194 while invoking one of the smp_call_function*()
5197 scftorture.verbose= [KNL]
5198 Enable additional printk() statements.
5200 scftorture.weight_single= [KNL]
5201 The probability weighting to use for the
5202 smp_call_function_single() function with a zero
5203 "wait" parameter. A value of -1 selects the
5204 default if all other weights are -1. However,
5205 if at least one weight has some other value, a
5206 value of -1 will instead select a weight of zero.
5208 scftorture.weight_single_wait= [KNL]
5209 The probability weighting to use for the
5210 smp_call_function_single() function with a
5211 non-zero "wait" parameter. See weight_single.
5213 scftorture.weight_many= [KNL]
5214 The probability weighting to use for the
5215 smp_call_function_many() function with a zero
5216 "wait" parameter. See weight_single.
5217 Note well that setting a high probability for
5218 this weighting can place serious IPI load
5221 scftorture.weight_many_wait= [KNL]
5222 The probability weighting to use for the
5223 smp_call_function_many() function with a
5224 non-zero "wait" parameter. See weight_single
5227 scftorture.weight_all= [KNL]
5228 The probability weighting to use for the
5229 smp_call_function_all() function with a zero
5230 "wait" parameter. See weight_single and
5233 scftorture.weight_all_wait= [KNL]
5234 The probability weighting to use for the
5235 smp_call_function_all() function with a
5236 non-zero "wait" parameter. See weight_single
5239 skew_tick= [KNL] Offset the periodic timer tick per cpu to mitigate
5240 xtime_lock contention on larger systems, and/or RCU lock
5241 contention on all systems with CONFIG_MAXSMP set.
5242 Format: { "0" | "1" }
5243 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
5245 Note: increases power consumption, thus should only be
5246 enabled if running jitter sensitive (HPC/RT) workloads.
5248 security= [SECURITY] Choose a legacy "major" security module to
5249 enable at boot. This has been deprecated by the
5252 selinux= [SELINUX] Disable or enable SELinux at boot time.
5253 Format: { "0" | "1" }
5254 See security/selinux/Kconfig help text.
5259 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
5260 Format: { "0" | "1" }
5261 See security/apparmor/Kconfig help text
5264 Default value is set via kernel config option.
5266 serialnumber [BUGS=X86-32]
5269 Maximal number of shapers.
5277 Enable merging of slabs with similar size when the
5278 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
5281 Disable merging of slabs with similar size. May be
5282 necessary if there is some reason to distinguish
5283 allocs to different slabs, especially in hardened
5284 environments where the risk of heap overflows and
5285 layout control by attackers can usually be
5286 frustrated by disabling merging. This will reduce
5287 most of the exposure of a heap attack to a single
5288 cache (risks via metadata attacks are mostly
5289 unchanged). Debug options disable merging on their
5291 For more information see Documentation/vm/slub.rst.
5293 slab_max_order= [MM, SLAB]
5294 Determines the maximum allowed order for slabs.
5295 A high setting may cause OOMs due to memory
5296 fragmentation. Defaults to 1 for systems with
5297 more than 32MB of RAM, 0 otherwise.
5299 slub_debug[=options[,slabs][;[options[,slabs]]...] [MM, SLUB]
5300 Enabling slub_debug allows one to determine the
5301 culprit if slab objects become corrupted. Enabling
5302 slub_debug can create guard zones around objects and
5303 may poison objects when not in use. Also tracks the
5304 last alloc / free. For more information see
5305 Documentation/vm/slub.rst.
5307 slub_max_order= [MM, SLUB]
5308 Determines the maximum allowed order for slabs.
5309 A high setting may cause OOMs due to memory
5310 fragmentation. For more information see
5311 Documentation/vm/slub.rst.
5313 slub_min_objects= [MM, SLUB]
5314 The minimum number of objects per slab. SLUB will
5315 increase the slab order up to slub_max_order to
5316 generate a sufficiently large slab able to contain
5317 the number of objects indicated. The higher the number
5318 of objects the smaller the overhead of tracking slabs
5319 and the less frequently locks need to be acquired.
5320 For more information see Documentation/vm/slub.rst.
5322 slub_min_order= [MM, SLUB]
5323 Determines the minimum page order for slabs. Must be
5324 lower than slub_max_order.
5325 For more information see Documentation/vm/slub.rst.
5327 slub_merge [MM, SLUB]
5328 Same with slab_merge.
5330 slub_nomerge [MM, SLUB]
5331 Same with slab_nomerge. This is supported for legacy.
5332 See slab_nomerge for more information.
5335 Format: <io1>[,<io2>[,...,<io8>]]
5337 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
5338 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
5339 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
5340 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
5341 smsc-ircc2.ircc_irq= [HW] IRQ line
5342 smsc-ircc2.ircc_dma= [HW] DMA channel
5343 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
5344 0: Toshiba Satellite 1800 (GP data pin select)
5345 1: Fast pin select (default)
5348 smt [KNL,S390] Set the maximum number of threads (logical
5349 CPUs) to use per physical CPU on systems capable of
5350 symmetric multithreading (SMT). Will be capped to the
5351 actual hardware limit.
5353 Default: -1 (no limit)
5356 [KNL] Should the soft-lockup detector generate panics.
5359 A value of 1 instructs the soft-lockup detector
5360 to panic the machine when a soft-lockup occurs. It is
5361 also controlled by the kernel.softlockup_panic sysctl
5362 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
5363 respective build-time switch to that functionality.
5365 softlockup_all_cpu_backtrace=
5366 [KNL] Should the soft-lockup detector generate
5367 backtraces on all cpus.
5370 sonypi.*= [HW] Sony Programmable I/O Control Device driver
5371 See Documentation/admin-guide/laptops/sonypi.rst
5373 spectre_v2= [X86] Control mitigation of Spectre variant 2
5374 (indirect branch speculation) vulnerability.
5375 The default operation protects the kernel from
5378 on - unconditionally enable, implies
5380 off - unconditionally disable, implies
5382 auto - kernel detects whether your CPU model is
5385 Selecting 'on' will, and 'auto' may, choose a
5386 mitigation method at run time according to the
5387 CPU, the available microcode, the setting of the
5388 CONFIG_RETPOLINE configuration option, and the
5389 compiler with which the kernel was built.
5391 Selecting 'on' will also enable the mitigation
5392 against user space to user space task attacks.
5394 Selecting 'off' will disable both the kernel and
5395 the user space protections.
5397 Specific mitigations can also be selected manually:
5399 retpoline - replace indirect branches
5400 retpoline,generic - Retpolines
5401 retpoline,lfence - LFENCE; indirect branch
5402 retpoline,amd - alias for retpoline,lfence
5403 eibrs - enhanced IBRS
5404 eibrs,retpoline - enhanced IBRS + Retpolines
5405 eibrs,lfence - enhanced IBRS + LFENCE
5406 ibrs - use IBRS to protect kernel
5408 Not specifying this option is equivalent to
5412 [X86] Control mitigation of Spectre variant 2
5413 (indirect branch speculation) vulnerability between
5416 on - Unconditionally enable mitigations. Is
5417 enforced by spectre_v2=on
5419 off - Unconditionally disable mitigations. Is
5420 enforced by spectre_v2=off
5422 prctl - Indirect branch speculation is enabled,
5423 but mitigation can be enabled via prctl
5424 per thread. The mitigation control state
5425 is inherited on fork.
5428 - Like "prctl" above, but only STIBP is
5429 controlled per thread. IBPB is issued
5430 always when switching between different user
5434 - Same as "prctl" above, but all seccomp
5435 threads will enable the mitigation unless
5436 they explicitly opt out.
5439 - Like "seccomp" above, but only STIBP is
5440 controlled per thread. IBPB is issued
5441 always when switching between different
5442 user space processes.
5444 auto - Kernel selects the mitigation depending on
5445 the available CPU features and vulnerability.
5448 If CONFIG_SECCOMP=y then "seccomp", otherwise "prctl"
5450 Not specifying this option is equivalent to
5451 spectre_v2_user=auto.
5453 spec_rstack_overflow=
5454 [X86] Control RAS overflow mitigation on AMD Zen CPUs
5456 off - Disable mitigation
5457 microcode - Enable microcode mitigation only
5458 safe-ret - Enable sw-only safe RET mitigation (default)
5459 ibpb - Enable mitigation by issuing IBPB on
5461 ibpb-vmexit - Issue IBPB only on VMEXIT
5462 (cloud-specific mitigation)
5464 spec_store_bypass_disable=
5465 [HW] Control Speculative Store Bypass (SSB) Disable mitigation
5466 (Speculative Store Bypass vulnerability)
5468 Certain CPUs are vulnerable to an exploit against a
5469 a common industry wide performance optimization known
5470 as "Speculative Store Bypass" in which recent stores
5471 to the same memory location may not be observed by
5472 later loads during speculative execution. The idea
5473 is that such stores are unlikely and that they can
5474 be detected prior to instruction retirement at the
5475 end of a particular speculation execution window.
5477 In vulnerable processors, the speculatively forwarded
5478 store can be used in a cache side channel attack, for
5479 example to read memory to which the attacker does not
5480 directly have access (e.g. inside sandboxed code).
5482 This parameter controls whether the Speculative Store
5483 Bypass optimization is used.
5485 On x86 the options are:
5487 on - Unconditionally disable Speculative Store Bypass
5488 off - Unconditionally enable Speculative Store Bypass
5489 auto - Kernel detects whether the CPU model contains an
5490 implementation of Speculative Store Bypass and
5491 picks the most appropriate mitigation. If the
5492 CPU is not vulnerable, "off" is selected. If the
5493 CPU is vulnerable the default mitigation is
5494 architecture and Kconfig dependent. See below.
5495 prctl - Control Speculative Store Bypass per thread
5496 via prctl. Speculative Store Bypass is enabled
5497 for a process by default. The state of the control
5498 is inherited on fork.
5499 seccomp - Same as "prctl" above, but all seccomp threads
5500 will disable SSB unless they explicitly opt out.
5502 Default mitigations:
5503 X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
5505 On powerpc the options are:
5507 on,auto - On Power8 and Power9 insert a store-forwarding
5508 barrier on kernel entry and exit. On Power7
5509 perform a software flush on kernel entry and
5513 Not specifying this option is equivalent to
5514 spec_store_bypass_disable=auto.
5516 spia_io_base= [HW,MTD]
5522 [X86] Enable split lock detection or bus lock detection
5524 When enabled (and if hardware support is present), atomic
5525 instructions that access data across cache line
5526 boundaries will result in an alignment check exception
5527 for split lock detection or a debug exception for
5532 warn - the kernel will emit rate-limited warnings
5533 about applications triggering the #AC
5534 exception or the #DB exception. This mode is
5535 the default on CPUs that support split lock
5536 detection or bus lock detection. Default
5537 behavior is by #AC if both features are
5538 enabled in hardware.
5540 fatal - the kernel will send SIGBUS to applications
5541 that trigger the #AC exception or the #DB
5542 exception. Default behavior is by #AC if
5543 both features are enabled in hardware.
5546 Set system wide rate limit to N bus locks
5547 per second for bus lock detection.
5550 N/A for split lock detection.
5553 If an #AC exception is hit in the kernel or in
5554 firmware (i.e. not while executing in user mode)
5555 the kernel will oops in either "warn" or "fatal"
5558 #DB exception for bus lock is triggered only when
5562 Control the Special Register Buffer Data Sampling
5565 Certain CPUs are vulnerable to an MDS-like
5566 exploit which can leak bits from the random
5569 By default, this issue is mitigated by
5570 microcode. However, the microcode fix can cause
5571 the RDRAND and RDSEED instructions to become
5572 much slower. Among other effects, this will
5573 result in reduced throughput from /dev/urandom.
5575 The microcode mitigation can be disabled with
5576 the following option:
5578 off: Disable mitigation and remove
5579 performance impact to RDRAND and RDSEED
5581 srcutree.counter_wrap_check [KNL]
5582 Specifies how frequently to check for
5583 grace-period sequence counter wrap for the
5584 srcu_data structure's ->srcu_gp_seq_needed field.
5585 The greater the number of bits set in this kernel
5586 parameter, the less frequently counter wrap will
5587 be checked for. Note that the bottom two bits
5590 srcutree.exp_holdoff [KNL]
5591 Specifies how many nanoseconds must elapse
5592 since the end of the last SRCU grace period for
5593 a given srcu_struct until the next normal SRCU
5594 grace period will be considered for automatic
5595 expediting. Set to zero to disable automatic
5599 Speculative Store Bypass Disable control
5601 On CPUs that are vulnerable to the Speculative
5602 Store Bypass vulnerability and offer a
5603 firmware based mitigation, this parameter
5604 indicates how the mitigation should be used:
5606 force-on: Unconditionally enable mitigation for
5607 for both kernel and userspace
5608 force-off: Unconditionally disable mitigation for
5609 for both kernel and userspace
5610 kernel: Always enable mitigation in the
5611 kernel, and offer a prctl interface
5612 to allow userspace to register its
5613 interest in being mitigated too.
5615 stack_guard_gap= [MM]
5616 override the default stack gap protection. The value
5617 is in page units and it defines how many pages prior
5618 to (for stacks growing down) resp. after (for stacks
5619 growing up) the main stack are reserved for no other
5620 mapping. Default value is 256 pages.
5622 stack_depot_disable= [KNL]
5623 Setting this to true through kernel command line will
5624 disable the stack depot thereby saving the static memory
5625 consumed by the stack hash table. By default this is set
5629 Enabled the stack tracer on boot up.
5631 stacktrace_filter=[function-list]
5632 [FTRACE] Limit the functions that the stack tracer
5633 will trace at boot up. function-list is a comma-separated
5634 list of functions. This list can be changed at run
5635 time by the stack_trace_filter file in the debugfs
5636 tracing directory. Note, this enables stack tracing
5637 and the stacktrace above is not needed.
5641 Set the STI (builtin display/keyboard on the HP-PARISC
5642 machines) console (graphic card) which should be used
5643 as the initial boot-console.
5644 See also comment in drivers/video/console/sticore.c.
5647 See comment in drivers/video/console/sticore.c.
5650 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
5652 sunrpc.min_resvport=
5653 sunrpc.max_resvport=
5655 SunRPC servers often require that client requests
5656 originate from a privileged port (i.e. a port in the
5657 range 0 < portnr < 1024).
5658 An administrator who wishes to reserve some of these
5659 ports for other uses may adjust the range that the
5660 kernel's sunrpc client considers to be privileged
5661 using these two parameters to set the minimum and
5662 maximum port values.
5664 sunrpc.svc_rpc_per_connection_limit=
5666 Limit the number of requests that the server will
5667 process in parallel from a single connection.
5668 The default value is 0 (no limit).
5672 Control how the NFS server code allocates CPUs to
5673 service thread pools. Depending on how many NICs
5674 you have and where their interrupts are bound, this
5675 option will affect which CPUs will do NFS serving.
5676 Note: this parameter cannot be changed while the
5677 NFS server is running.
5679 auto the server chooses an appropriate mode
5680 automatically using heuristics
5681 global a single global pool contains all CPUs
5682 percpu one pool for each CPU
5683 pernode one pool for each NUMA node (equivalent
5684 to global on non-NUMA machines)
5686 sunrpc.tcp_slot_table_entries=
5687 sunrpc.udp_slot_table_entries=
5689 Sets the upper limit on the number of simultaneous
5690 RPC calls that can be sent from the client to a
5691 server. Increasing these values may allow you to
5692 improve throughput, but will also increase the
5693 amount of memory reserved for use by the client.
5695 suspend.pm_test_delay=
5697 Sets the number of seconds to remain in a suspend test
5698 mode before resuming the system (see
5699 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
5700 is set. Default value is 5.
5703 Format: { on | off | y | n | 1 | 0 }
5704 This parameter controls use of the Protected
5705 Execution Facility on pSeries.
5708 [KNL] Enable accounting of swap in memory resource
5709 controller if no parameter or 1 is given or disable
5710 it if 0 is given (See Documentation/admin-guide/cgroup-v1/memory.rst)
5712 swiotlb= [ARM,IA-64,PPC,MIPS,X86]
5713 Format: { <int> | force | noforce }
5714 <int> -- Number of I/O TLB slabs
5715 force -- force using of bounce buffers even if they
5716 wouldn't be automatically used by the kernel
5717 noforce -- Never use bounce buffers (for debugging)
5722 Set a sysctl parameter, right before loading the init
5723 process, as if the value was written to the respective
5724 /proc/sys/... file. Both '.' and '/' are recognized as
5725 separators. Unrecognized parameters and invalid values
5726 are reported in the kernel log. Sysctls registered
5727 later by a loaded module cannot be set this way.
5728 Example: sysctl.vm.swappiness=40
5730 sysfs.deprecated=0|1 [KNL]
5731 Enable/disable old style sysfs layout for old udev
5732 on older distributions. When this option is enabled
5733 very new udev will not work anymore. When this option
5734 is disabled (or CONFIG_SYSFS_DEPRECATED not compiled)
5735 in older udev will not work anymore.
5736 Default depends on CONFIG_SYSFS_DEPRECATED_V2 set in
5737 the kernel configuration.
5739 sysrq_always_enabled
5741 Ignore sysrq setting - this boot parameter will
5742 neutralize any effect of /proc/sys/kernel/sysrq.
5743 Useful for debugging.
5745 tcpmhash_entries= [KNL,NET]
5746 Set the number of tcp_metrics_hash slots.
5747 Default value is 8192 or 16384 depending on total
5748 ram pages. This is used to specify the TCP metrics
5749 cache size. See Documentation/networking/ip-sysctl.rst
5750 "tcp_no_metrics_save" section for more details.
5754 test_suspend= [SUSPEND][,N]
5755 Specify "mem" (for Suspend-to-RAM) or "standby" (for
5756 standby suspend) or "freeze" (for suspend type freeze)
5757 as the system sleep state during system startup with
5758 the optional capability to repeat N number of times.
5759 The system is woken from this state using a
5760 wakeup-capable RTC alarm.
5762 thash_entries= [KNL,NET]
5763 Set number of hash buckets for TCP connection
5765 thermal.act= [HW,ACPI]
5766 -1: disable all active trip points in all thermal zones
5767 <degrees C>: override all lowest active trip points
5769 thermal.crt= [HW,ACPI]
5770 -1: disable all critical trip points in all thermal zones
5771 <degrees C>: override all critical trip points
5773 thermal.off= [HW,ACPI]
5774 1: disable ACPI thermal control
5776 thermal.psv= [HW,ACPI]
5777 -1: disable all passive trip points
5778 <degrees C>: override all passive trip points to this
5781 thermal.tzp= [HW,ACPI]
5782 Specify global default ACPI thermal zone polling rate
5783 <deci-seconds>: poll all this frequency
5784 0: no polling (default)
5787 Force threading of all interrupt handlers except those
5788 marked explicitly IRQF_NO_THREAD.
5792 Specify if the kernel should make use of the cpu
5793 topology information if the hardware supports this.
5794 The scheduler will make use of this information and
5795 e.g. base its process migration decisions on it.
5798 topology_updates= [KNL, PPC, NUMA]
5800 Specify if the kernel should ignore (off)
5801 topology updates sent by the hypervisor to this
5804 torture.disable_onoff_at_boot= [KNL]
5805 Prevent the CPU-hotplug component of torturing
5806 until after init has spawned.
5808 torture.ftrace_dump_at_shutdown= [KNL]
5809 Dump the ftrace buffer at torture-test shutdown,
5810 even if there were no errors. This can be a
5811 very costly operation when many torture tests
5812 are running concurrently, especially on systems
5813 with rotating-rust storage.
5815 torture.verbose_sleep_frequency= [KNL]
5816 Specifies how many verbose printk()s should be
5817 emitted between each sleep. The default of zero
5818 disables verbose-printk() sleeping.
5820 torture.verbose_sleep_duration= [KNL]
5821 Duration of each verbose-printk() sleep in jiffies.
5825 tpm_suspend_pcr=[HW,TPM]
5826 Format: integer pcr id
5827 Specify that at suspend time, the tpm driver
5828 should extend the specified pcr with zeros,
5829 as a workaround for some chips which fail to
5830 flush the last written pcr on TPM_SaveState.
5831 This will guarantee that all the other pcrs
5834 trace_buf_size=nn[KMG]
5835 [FTRACE] will set tracing buffer size on each cpu.
5837 trace_event=[event-list]
5838 [FTRACE] Set and start specified trace events in order
5839 to facilitate early boot debugging. The event-list is a
5840 comma-separated list of trace events to enable. See
5841 also Documentation/trace/events.rst
5843 trace_options=[option-list]
5844 [FTRACE] Enable or disable tracer options at boot.
5845 The option-list is a comma delimited list of options
5846 that can be enabled or disabled just as if you were
5847 to echo the option name into
5849 /sys/kernel/debug/tracing/trace_options
5851 For example, to enable stacktrace option (to dump the
5852 stack trace of each event), add to the command line:
5854 trace_options=stacktrace
5856 See also Documentation/trace/ftrace.rst "trace options"
5860 Have the tracepoints sent to printk as well as the
5861 tracing ring buffer. This is useful for early boot up
5862 where the system hangs or reboots and does not give the
5863 option for reading the tracing buffer or performing a
5864 ftrace_dump_on_oops.
5866 To turn off having tracepoints sent to printk,
5867 echo 0 > /proc/sys/kernel/tracepoint_printk
5868 Note, echoing 1 into this file without the
5869 tracepoint_printk kernel cmdline option has no effect.
5871 The tp_printk_stop_on_boot (see below) can also be used
5872 to stop the printing of events to console at
5877 Having tracepoints sent to printk() and activating high
5878 frequency tracepoints such as irq or sched, can cause
5879 the system to live lock.
5881 tp_printk_stop_on_boot[FTRACE]
5882 When tp_printk (above) is set, it can cause a lot of noise
5883 on the console. It may be useful to only include the
5884 printing of events during boot up, as user space may
5885 make the system inoperable.
5887 This command line option will stop the printing of events
5888 to console at the late_initcall_sync() time frame.
5891 [FTRACE] enable this option to disable tracing when a
5892 warning is hit. This turns off "tracing_on". Tracing can
5893 be enabled again by echoing '1' into the "tracing_on"
5894 file located in /sys/kernel/debug/tracing/
5896 This option is useful, as it disables the trace before
5897 the WARNING dump is called, which prevents the trace to
5898 be filled with content caused by the warning output.
5900 This option can also be set at run time via the sysctl
5901 option: kernel/traceoff_on_warning
5903 transparent_hugepage=
5905 Format: [always|madvise|never]
5906 Can be used to control the default behavior of the system
5907 with respect to transparent hugepages.
5908 See Documentation/admin-guide/mm/transhuge.rst
5911 trusted.source= [KEYS]
5913 This parameter identifies the trust source as a backend
5914 for trusted keys implementation. Supported trust
5918 If not specified then it defaults to iterating through
5919 the trust source list starting with TPM and assigns the
5920 first trust source as a backend which is initialized
5921 successfully during iteration.
5923 tsc= Disable clocksource stability checks for TSC.
5925 [x86] reliable: mark tsc clocksource as reliable, this
5926 disables clocksource verification at runtime, as well
5927 as the stability checks done at bootup. Used to enable
5928 high-resolution timer mode on older hardware, and in
5929 virtualized environment.
5930 [x86] noirqtime: Do not use TSC to do irq accounting.
5931 Used to run time disable IRQ_TIME_ACCOUNTING on any
5932 platforms where RDTSC is slow and this accounting
5934 [x86] unstable: mark the TSC clocksource as unstable, this
5935 marks the TSC unconditionally unstable at bootup and
5936 avoids any further wobbles once the TSC watchdog notices.
5937 [x86] nowatchdog: disable clocksource watchdog. Used
5938 in situations with strict latency requirements (where
5939 interruptions from clocksource watchdog are not
5942 tsc_early_khz= [X86] Skip early TSC calibration and use the given
5943 value instead. Useful when the early TSC frequency discovery
5944 procedure is not reliable, such as on overclocked systems
5945 with CPUID.16h support and partial CPUID.15h support.
5946 Format: <unsigned int>
5948 tsx= [X86] Control Transactional Synchronization
5949 Extensions (TSX) feature in Intel processors that
5950 support TSX control.
5952 This parameter controls the TSX feature. The options are:
5954 on - Enable TSX on the system. Although there are
5955 mitigations for all known security vulnerabilities,
5956 TSX has been known to be an accelerator for
5957 several previous speculation-related CVEs, and
5958 so there may be unknown security risks associated
5959 with leaving it enabled.
5961 off - Disable TSX on the system. (Note that this
5962 option takes effect only on newer CPUs which are
5963 not vulnerable to MDS, i.e., have
5964 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
5965 the new IA32_TSX_CTRL MSR through a microcode
5966 update. This new MSR allows for the reliable
5967 deactivation of the TSX functionality.)
5969 auto - Disable TSX if X86_BUG_TAA is present,
5970 otherwise enable TSX on the system.
5972 Not specifying this option is equivalent to tsx=off.
5974 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
5977 tsx_async_abort= [X86,INTEL] Control mitigation for the TSX Async
5978 Abort (TAA) vulnerability.
5980 Similar to Micro-architectural Data Sampling (MDS)
5981 certain CPUs that support Transactional
5982 Synchronization Extensions (TSX) are vulnerable to an
5983 exploit against CPU internal buffers which can forward
5984 information to a disclosure gadget under certain
5987 In vulnerable processors, the speculatively forwarded
5988 data can be used in a cache side channel attack, to
5989 access data to which the attacker does not have direct
5992 This parameter controls the TAA mitigation. The
5995 full - Enable TAA mitigation on vulnerable CPUs
5998 full,nosmt - Enable TAA mitigation and disable SMT on
5999 vulnerable CPUs. If TSX is disabled, SMT
6000 is not disabled because CPU is not
6001 vulnerable to cross-thread TAA attacks.
6002 off - Unconditionally disable TAA mitigation
6004 On MDS-affected machines, tsx_async_abort=off can be
6005 prevented by an active MDS mitigation as both vulnerabilities
6006 are mitigated with the same mechanism so in order to disable
6007 this mitigation, you need to specify mds=off too.
6009 Not specifying this option is equivalent to
6010 tsx_async_abort=full. On CPUs which are MDS affected
6011 and deploy MDS mitigation, TAA mitigation is not
6012 required and doesn't provide any additional
6016 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6018 turbografx.map[2|3]= [HW,JOY]
6019 TurboGraFX parallel port interface
6021 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6022 See also Documentation/input/devices/joystick-parport.rst
6024 udbg-immortal [PPC] When debugging early kernel crashes that
6025 happen after console_init() and before a proper
6026 console driver takes over, this boot options might
6027 help "seeing" what's going on.
6029 uhash_entries= [KNL,NET]
6030 Set number of hash buckets for UDP/UDP-Lite connections
6033 [USB] Ignore overcurrent events (default N).
6034 Some badly-designed motherboards generate lots of
6035 bogus events, for ports that aren't wired to
6036 anything. Set this parameter to avoid log spamming.
6037 Note that genuine overcurrent events won't be
6041 [X86] Cause panic on unknown NMI.
6043 usbcore.authorized_default=
6044 [USB] Default USB device authorization:
6045 (default -1 = authorized except for wireless USB,
6046 0 = not authorized, 1 = authorized, 2 = authorized
6047 if device connected to internal port)
6049 usbcore.autosuspend=
6050 [USB] The autosuspend time delay (in seconds) used
6051 for newly-detected USB devices (default 2). This
6052 is the time required before an idle device will be
6053 autosuspended. Devices for which the delay is set
6054 to a negative value won't be autosuspended at all.
6056 usbcore.usbfs_snoop=
6057 [USB] Set to log all usbfs traffic (default 0 = off).
6059 usbcore.usbfs_snoop_max=
6060 [USB] Maximum number of bytes to snoop in each URB
6063 usbcore.blinkenlights=
6064 [USB] Set to cycle leds on hubs (default 0 = off).
6066 usbcore.old_scheme_first=
6067 [USB] Start with the old device initialization
6068 scheme (default 0 = off).
6070 usbcore.usbfs_memory_mb=
6071 [USB] Memory limit (in MB) for buffers allocated by
6072 usbfs (default = 16, 0 = max = 2047).
6074 usbcore.use_both_schemes=
6075 [USB] Try the other device initialization scheme
6076 if the first one fails (default 1 = enabled).
6078 usbcore.initial_descriptor_timeout=
6079 [USB] Specifies timeout for the initial 64-byte
6080 USB_REQ_GET_DESCRIPTOR request in milliseconds
6081 (default 5000 = 5.0 seconds).
6083 usbcore.nousb [USB] Disable the USB subsystem
6086 [USB] A list of quirk entries to augment the built-in
6087 usb core quirk list. List entries are separated by
6088 commas. Each entry has the form
6089 VendorID:ProductID:Flags. The IDs are 4-digit hex
6090 numbers and Flags is a set of letters. Each letter
6091 will change the built-in quirk; setting it if it is
6092 clear and clearing it if it is set. The letters have
6093 the following meanings:
6094 a = USB_QUIRK_STRING_FETCH_255 (string
6095 descriptors must not be fetched using
6097 b = USB_QUIRK_RESET_RESUME (device can't resume
6098 correctly so reset it instead);
6099 c = USB_QUIRK_NO_SET_INTF (device can't handle
6100 Set-Interface requests);
6101 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
6102 handle its Configuration or Interface
6104 e = USB_QUIRK_RESET (device can't be reset
6105 (e.g morph devices), don't use reset);
6106 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
6107 more interface descriptions than the
6108 bNumInterfaces count, and can't handle
6109 talking to these interfaces);
6110 g = USB_QUIRK_DELAY_INIT (device needs a pause
6111 during initialization, after we read
6112 the device descriptor);
6113 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
6114 high speed and super speed interrupt
6115 endpoints, the USB 2.0 and USB 3.0 spec
6116 require the interval in microframes (1
6117 microframe = 125 microseconds) to be
6118 calculated as interval = 2 ^
6120 Devices with this quirk report their
6121 bInterval as the result of this
6122 calculation instead of the exponent
6123 variable used in the calculation);
6124 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
6125 handle device_qualifier descriptor
6127 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
6128 generates spurious wakeup, ignore
6129 remote wakeup capability);
6130 k = USB_QUIRK_NO_LPM (device can't handle Link
6132 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
6133 (Device reports its bInterval as linear
6134 frames instead of the USB 2.0
6136 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
6137 to be disconnected before suspend to
6138 prevent spurious wakeup);
6139 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
6140 pause after every control message);
6141 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
6142 delay after resetting its port);
6143 Example: quirks=0781:5580:bk,0a5c:5834:gij
6146 [USBHID] The interval which mice are to be polled at.
6149 [USBHID] The interval which joysticks are to be polled at.
6152 [USBHID] The interval which keyboards are to be polled at.
6154 usb-storage.delay_use=
6155 [UMS] The delay in seconds before a new device is
6156 scanned for Logical Units (default 1).
6159 [UMS] A list of quirks entries to supplement or
6160 override the built-in unusual_devs list. List
6161 entries are separated by commas. Each entry has
6162 the form VID:PID:Flags where VID and PID are Vendor
6163 and Product ID values (4-digit hex numbers) and
6164 Flags is a set of characters, each corresponding
6165 to a common usb-storage quirk flag as follows:
6166 a = SANE_SENSE (collect more than 18 bytes
6167 of sense data, not on uas);
6168 b = BAD_SENSE (don't collect more than 18
6169 bytes of sense data, not on uas);
6170 c = FIX_CAPACITY (decrease the reported
6171 device capacity by one sector);
6172 d = NO_READ_DISC_INFO (don't use
6173 READ_DISC_INFO command, not on uas);
6174 e = NO_READ_CAPACITY_16 (don't use
6175 READ_CAPACITY_16 command);
6176 f = NO_REPORT_OPCODES (don't use report opcodes
6178 g = MAX_SECTORS_240 (don't transfer more than
6179 240 sectors at a time, uas only);
6180 h = CAPACITY_HEURISTICS (decrease the
6181 reported device capacity by one
6182 sector if the number is odd);
6183 i = IGNORE_DEVICE (don't bind to this
6185 j = NO_REPORT_LUNS (don't use report luns
6187 k = NO_SAME (do not use WRITE_SAME, uas only)
6188 l = NOT_LOCKABLE (don't try to lock and
6189 unlock ejectable media, not on uas);
6190 m = MAX_SECTORS_64 (don't transfer more
6191 than 64 sectors = 32 KB at a time,
6193 n = INITIAL_READ10 (force a retry of the
6194 initial READ(10) command, not on uas);
6195 o = CAPACITY_OK (accept the capacity
6196 reported by the device, not on uas);
6197 p = WRITE_CACHE (the device cache is ON
6198 by default, not on uas);
6199 r = IGNORE_RESIDUE (the device reports
6200 bogus residue values, not on uas);
6201 s = SINGLE_LUN (the device has only one
6203 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
6204 commands, uas only);
6205 u = IGNORE_UAS (don't bind to the uas driver);
6206 w = NO_WP_DETECT (don't test whether the
6207 medium is write-protected).
6208 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
6209 even if the device claims no cache,
6211 Example: quirks=0419:aaf5:rl,0421:0433:rc
6213 user_debug= [KNL,ARM]
6215 See arch/arm/Kconfig.debug help text.
6216 1 - undefined instruction events
6218 4 - invalid data aborts
6221 Example: user_debug=31
6224 [X86] Flags controlling user PTE allocations.
6226 nohigh = do not allocate PTE pages in
6227 HIGHMEM regardless of setting
6231 On X86_32, this is an alias for vdso32=. Otherwise:
6233 vdso=1: enable VDSO (the default)
6234 vdso=0: disable VDSO mapping
6236 vdso32= [X86] Control the 32-bit vDSO
6237 vdso32=1: enable 32-bit VDSO
6238 vdso32=0 or vdso32=2: disable 32-bit VDSO
6240 See the help text for CONFIG_COMPAT_VDSO for more
6241 details. If CONFIG_COMPAT_VDSO is set, the default is
6242 vdso32=0; otherwise, the default is vdso32=1.
6244 For compatibility with older kernels, vdso32=2 is an
6247 Try vdso32=0 if you encounter an error that says:
6248 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
6251 vector=percpu: enable percpu vector domain
6253 video= [FB] Frame buffer configuration
6254 See Documentation/fb/modedb.rst.
6256 video.brightness_switch_enabled= [0,1]
6257 If set to 1, on receiving an ACPI notify event
6258 generated by hotkey, video driver will adjust brightness
6259 level and then send out the event to user space through
6260 the allocated input device; If set to 0, video driver
6261 will only send out the event without touching backlight
6266 [VMMIO] Memory mapped virtio (platform) device.
6268 <size>@<baseaddr>:<irq>[:<id>]
6270 <size> := size (can use standard suffixes
6272 <baseaddr> := physical base address
6273 <irq> := interrupt number (as passed to
6275 <id> := (optional) platform device id
6277 virtio_mmio.device=1K@0x100b0000:48:7
6279 Can be used multiple times for multiple devices.
6281 vga= [BOOT,X86-32] Select a particular video mode
6282 See Documentation/x86/boot.rst and
6283 Documentation/admin-guide/svga.rst.
6284 Use vga=ask for menu.
6285 This is actually a boot loader parameter; the value is
6286 passed to the kernel using a special protocol.
6288 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
6289 May slow down system boot speed, especially when
6290 enabled on systems with a large amount of memory.
6291 All options are enabled by default, and this
6292 interface is meant to allow for selectively
6293 enabling or disabling specific virtual memory
6296 Available options are:
6297 P Enable page structure init time poisoning
6298 - Disable all of the above options
6300 vmalloc=nn[KMG] [KNL,BOOT] Forces the vmalloc area to have an exact
6301 size of <nn>. This can be used to increase the
6302 minimum size (128MB on x86). It can also be used to
6303 decrease the size and leave more room for directly
6306 vmcp_cma=nn[MG] [KNL,S390]
6307 Sets the memory size reserved for contiguous memory
6308 allocations for the vmcp device driver.
6310 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
6313 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
6316 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
6320 Controls the behavior of vsyscalls (i.e. calls to
6321 fixed addresses of 0xffffffffff600x00 from legacy
6322 code). Most statically-linked binaries and older
6323 versions of glibc use these calls. Because these
6324 functions are at fixed addresses, they make nice
6325 targets for exploits that can control RIP.
6327 emulate [default] Vsyscalls turn into traps and are
6328 emulated reasonably safely. The vsyscall
6331 xonly Vsyscalls turn into traps and are
6332 emulated reasonably safely. The vsyscall
6333 page is not readable.
6335 none Vsyscalls don't work at all. This makes
6336 them quite hard to use for exploits but
6337 might break your system.
6339 vt.color= [VT] Default text color.
6340 Format: 0xYX, X = foreground, Y = background.
6341 Default: 0x07 = light gray on black.
6343 vt.cur_default= [VT] Default cursor shape.
6344 Format: 0xCCBBAA, where AA, BB, and CC are the same as
6345 the parameters of the <Esc>[?A;B;Cc escape sequence;
6346 see VGA-softcursor.txt. Default: 2 = underline.
6348 vt.default_blu= [VT]
6349 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
6350 Change the default blue palette of the console.
6351 This is a 16-member array composed of values
6354 vt.default_grn= [VT]
6355 Format: <green0>,<green1>,<green2>,...,<green15>
6356 Change the default green palette of the console.
6357 This is a 16-member array composed of values
6360 vt.default_red= [VT]
6361 Format: <red0>,<red1>,<red2>,...,<red15>
6362 Change the default red palette of the console.
6363 This is a 16-member array composed of values
6369 Set system-wide default UTF-8 mode for all tty's.
6370 Default is 1, i.e. UTF-8 mode is enabled for all
6371 newly opened terminals.
6373 vt.global_cursor_default=
6376 Set system-wide default for whether a cursor
6377 is shown on new VTs. Default is -1,
6378 i.e. cursors will be created by default unless
6379 overridden by individual drivers. 0 will hide
6380 cursors, 1 will display them.
6382 vt.italic= [VT] Default color for italic text; 0-15.
6385 vt.underline= [VT] Default color for underlined text; 0-15.
6388 watchdog timers [HW,WDT] For information on watchdog timers,
6389 see Documentation/watchdog/watchdog-parameters.rst
6390 or other driver-specific files in the
6391 Documentation/watchdog/ directory.
6395 Set the hard lockup detector stall duration
6396 threshold in seconds. The soft lockup detector
6397 threshold is set to twice the value. A value of 0
6398 disables both lockup detectors. Default is 10
6401 workqueue.watchdog_thresh=
6402 If CONFIG_WQ_WATCHDOG is configured, workqueue can
6403 warn stall conditions and dump internal state to
6404 help debugging. 0 disables workqueue stall
6405 detection; otherwise, it's the stall threshold
6406 duration in seconds. The default value is 30 and
6407 it can be updated at runtime by writing to the
6408 corresponding sysfs file.
6410 workqueue.disable_numa
6411 By default, all work items queued to unbound
6412 workqueues are affine to the NUMA nodes they're
6413 issued on, which results in better behavior in
6414 general. If NUMA affinity needs to be disabled for
6415 whatever reason, this option can be used. Note
6416 that this also can be controlled per-workqueue for
6417 workqueues visible under /sys/bus/workqueue/.
6419 workqueue.power_efficient
6420 Per-cpu workqueues are generally preferred because
6421 they show better performance thanks to cache
6422 locality; unfortunately, per-cpu workqueues tend to
6423 be more power hungry than unbound workqueues.
6425 Enabling this makes the per-cpu workqueues which
6426 were observed to contribute significantly to power
6427 consumption unbound, leading to measurably lower
6428 power usage at the cost of small performance
6431 The default value of this parameter is determined by
6432 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
6434 workqueue.debug_force_rr_cpu
6435 Workqueue used to implicitly guarantee that work
6436 items queued without explicit CPU specified are put
6437 on the local CPU. This guarantee is no longer true
6438 and while local CPU is still preferred work items
6439 may be put on foreign CPUs. This debug option
6440 forces round-robin CPU selection to flush out
6441 usages which depend on the now broken guarantee.
6442 When enabled, memory and cache locality will be
6445 x2apic_phys [X86-64,APIC] Use x2apic physical mode instead of
6446 default x2apic cluster mode on platforms
6449 xen_512gb_limit [KNL,X86-64,XEN]
6450 Restricts the kernel running paravirtualized under Xen
6451 to use only up to 512 GB of RAM. The reason to do so is
6452 crash analysis tools and Xen tools for doing domain
6453 save/restore/migration must be enabled to handle larger
6456 xen_emul_unplug= [HW,X86,XEN]
6457 Unplug Xen emulated devices
6458 Format: [unplug0,][unplug1]
6459 ide-disks -- unplug primary master IDE devices
6460 aux-ide-disks -- unplug non-primary-master IDE devices
6461 nics -- unplug network devices
6462 all -- unplug all emulated devices (NICs and IDE disks)
6463 unnecessary -- unplugging emulated devices is
6464 unnecessary even if the host did not respond to
6466 never -- do not unplug even if version check succeeds
6468 xen_legacy_crash [X86,XEN]
6469 Crash from Xen panic notifier, without executing late
6470 panic() code such as dumping handler.
6472 xen_nopvspin [X86,XEN]
6473 Disables the qspinlock slowpath using Xen PV optimizations.
6474 This parameter is obsoleted by "nopvspin" parameter, which
6475 has equivalent effect for XEN platform.
6478 Disables the PV optimizations forcing the HVM guest to
6479 run as generic HVM guest with no PV drivers.
6480 This option is obsoleted by the "nopv" option, which
6481 has equivalent effect for XEN platform.
6483 xen_no_vector_callback
6484 [KNL,X86,XEN] Disable the vector callback for Xen
6485 event channel interrupts.
6487 xen_scrub_pages= [XEN]
6488 Boolean option to control scrubbing pages before giving them back
6489 to Xen, for use by other domains. Can be also changed at runtime
6490 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
6491 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
6493 xen_timer_slop= [X86-64,XEN]
6494 Set the timer slop (in nanoseconds) for the virtual Xen
6495 timers (default is 100000). This adjusts the minimum
6496 delta of virtualized Xen timers, where lower values
6497 improve timer resolution at the expense of processing
6498 more timer interrupts.
6500 xen.balloon_boot_timeout= [XEN]
6501 The time (in seconds) to wait before giving up to boot
6502 in case initial ballooning fails to free enough memory.
6503 Applies only when running as HVM or PVH guest and
6504 started with less memory configured than allowed at
6505 max. Default is 180.
6507 xen.event_eoi_delay= [XEN]
6508 How long to delay EOI handling in case of event
6509 storms (jiffies). Default is 10.
6511 xen.event_loop_timeout= [XEN]
6512 After which time (jiffies) the event handling loop
6513 should start to delay EOI handling. Default is 2.
6515 xen.fifo_events= [XEN]
6516 Boolean parameter to disable using fifo event handling
6517 even if available. Normally fifo event handling is
6518 preferred over the 2-level event handling, as it is
6519 fairer and the number of possible event channels is
6520 much higher. Default is on (use fifo events).
6522 nopv= [X86,XEN,KVM,HYPER_V,VMWARE]
6523 Disables the PV optimizations forcing the guest to run
6524 as generic guest with no PV drivers. Currently support
6525 XEN HVM, KVM, HYPER_V and VMWARE guest.
6527 nopvspin [X86,XEN,KVM]
6528 Disables the qspinlock slow path using PV optimizations
6529 which allow the hypervisor to 'idle' the guest on lock
6532 xirc2ps_cs= [NET,PCMCIA]
6534 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
6537 By default on POWER9 and above, the kernel will
6538 natively use the XIVE interrupt controller. This option
6539 allows the fallback firmware mode to be used:
6541 off Fallback to firmware control of XIVE interrupt
6542 controller on both pseries and powernv
6543 platforms. Only useful on POWER9 and above.
6545 xhci-hcd.quirks [USB,KNL]
6546 A hex value specifying bitmask with supplemental xhci
6547 host controller quirks. Meaning of each bit can be
6548 consulted in header drivers/usb/host/xhci.h.
6551 Format: { early | on | rw | ro | off }
6552 Controls if xmon debugger is enabled. Default is off.
6553 Passing only "xmon" is equivalent to "xmon=early".
6554 early Call xmon as early as possible on boot; xmon
6555 debugger is called from setup_arch().
6556 on xmon debugger hooks will be installed so xmon
6557 is only called on a kernel crash. Default mode,
6558 i.e. either "ro" or "rw" mode, is controlled
6559 with CONFIG_XMON_DEFAULT_RO_MODE.
6560 rw xmon debugger hooks will be installed so xmon
6561 is called only on a kernel crash, mode is write,
6562 meaning SPR registers, memory and, other data
6563 can be written using xmon commands.
6564 ro same as "rw" option above but SPR registers,
6565 memory, and other data can't be written using
6567 off xmon is disabled.